Archive fork in favor of upstream and DUMAS nodes
tests / test (push) Canceled after 0s
tests / windows-storage (push) Canceled after 0s

This commit is contained in:
2026-10-20 10:46:16 +00:00
parent 881dcbcfba
commit 3716b02a57
76 changed files with 12140 additions and 4410 deletions
+27
View File
@@ -56,6 +56,10 @@ examples/workflows/*
!examples/workflows/MiniMax_H3_Continuum_V39.json
!examples/workflows/MiniMax_H3_Continuum_V39.zip
!examples/MiniMax_H3_Continuum_V39.json
!examples/MiniMax_H3_Continuum_V39_Turbo_Sample.json
!examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.json
!examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.zip
!docs/TURBO_SAMPLE.md
!examples/MiniMax_H3_Continuum_V38X2.json
# Current loader migration instructions.
@@ -70,3 +74,26 @@ examples/workflows/*
# Keep the Registry payload aligned with its declared Manifest.
!docs/REVIEW_RESTORATION_REPAIR.md
write-h3-v39-reference-prompts/
# Current Issue #31 compatibility and update guidance.
!docs/ISSUE31_SAMPLING_SIGNATURES.md
# Current regeneration intent and automatic variation guide.
!docs/REGENERATE_INTENT_V396.md
# Current audited KJ Preview Override compatibility guide.
!docs/KJ_PREVIEW_OVERRIDE_V398.md
# Keep the condensed README and its moved text references together.
!docs/USER_GUIDE.md
!docs/README_HISTORY.md
# Japanese detailed references retained by the condensed README.
!docs/USER_GUIDE_JA.md
!docs/README_HISTORY_JA.md
# Current 3.9.9 update and verification notes.
!docs/RELEASE_NOTES_V399.md
# Current 3.9.10 repair/update notes.
!docs/RELEASE_NOTES_V3910.md
+5 -5
View File
@@ -7,7 +7,7 @@ permissions:
contents: read
concurrency:
group: publish-comfy-registry-v3.9.4
group: publish-comfy-registry-v3.9.10
cancel-in-progress: false
jobs:
@@ -15,10 +15,10 @@ jobs:
runs-on: ubuntu-latest
timeout-minutes: 15
steps:
- name: Check out v3.9.4 release
- name: Check out v3.9.10 release
uses: actions/checkout@v4
with:
ref: v3.9.4
ref: v3.9.10
persist-credentials: false
- name: Set up Python
uses: actions/setup-python@v5
@@ -27,14 +27,14 @@ jobs:
- name: Verify release identity and source hashes
shell: bash
run: |
test "$(git describe --tags --exact-match HEAD)" = "v3.9.4"
test "$(git describe --tags --exact-match HEAD)" = "v3.9.10"
python - <<'PY'
import hashlib
import pathlib
import tomllib
root = pathlib.Path(".")
metadata = tomllib.loads((root / "pyproject.toml").read_text())
assert metadata["project"]["version"] == "3.9.4"
assert metadata["project"]["version"] == "3.9.10"
assert metadata["tool"]["comfy"]["PublisherId"] == "ukr8b3g-cmyk"
for name in ("REGISTRY_MANIFEST.sha256", "MANIFEST.sha256"):
rows = (root / name).read_text().splitlines()
+154
View File
@@ -0,0 +1,154 @@
name: Publish verified v3.9.10
on:
workflow_run:
workflows: [tests]
types: [completed]
permissions:
contents: read
concurrency:
group: publish-v3.9.10
cancel-in-progress: false
jobs:
tag:
if: github.event.workflow_run.conclusion == 'success' && github.event.workflow_run.head_branch == 'main' && github.event.workflow_run.head_repository.full_name == github.repository
runs-on: ubuntu-latest
permissions:
contents: write
outputs:
publish: ${{ steps.tag.outputs.publish }}
steps:
- uses: actions/checkout@v4
with:
ref: ${{ github.event.workflow_run.head_sha }}
- id: tag
env:
GH_TOKEN: ${{ github.token }}
RELEASE_SHA: ${{ github.event.workflow_run.head_sha }}
run: |
test "$(python -c 'import tomllib;print(tomllib.load(open("pyproject.toml","rb"))["project"]["version"])')" = 3.9.10
test "$(gh api repos/$GITHUB_REPOSITORY/branches/main --jq .commit.sha)" = "$RELEASE_SHA"
if gh release view v3.9.10 --json isDraft --jq .isDraft 2>/dev/null | grep -qx false; then
echo 'publish=false' >> "$GITHUB_OUTPUT"
exit 0
fi
if EXISTING=$(gh api repos/$GITHUB_REPOSITORY/git/ref/tags/v3.9.10 --jq .object.sha 2>/dev/null); then
echo "Existing tag: $EXISTING"
test "$EXISTING" = "$RELEASE_SHA"
else
echo "Creating v3.9.10 at $RELEASE_SHA"
gh api repos/$GITHUB_REPOSITORY/git/refs -f ref=refs/tags/v3.9.10 -f sha="$RELEASE_SHA"
fi
echo 'publish=true' >> "$GITHUB_OUTPUT"
tag-tests:
needs: tag
if: needs.tag.outputs.publish == 'true'
strategy:
fail-fast: false
matrix:
os: [ubuntu-latest, windows-latest]
runs-on: ${{ matrix.os }}
steps:
- uses: actions/checkout@v4
with:
ref: v3.9.10
- uses: actions/setup-python@v5
with:
python-version: '3.11'
- name: Install CPU test dependencies
run: |
python -m pip install --index-url https://download.pytorch.org/whl/cpu torch
python -m pip install -r requirements.txt numpy safetensors pytest psutil opencv-python-headless Pillow aiohttp
- name: Compile tag
run: python -m compileall -q .
- name: Linux full regression on published tag
if: runner.os == 'Linux'
run: python -m pytest -q -p no:cacheprovider
- name: Windows storage regression on published tag
if: runner.os == 'Windows'
run: python -m pytest -q -p no:cacheprovider tests/test_temporary_storage.py tests/test_storage_api.py tests/test_p20_file_backed_tensor_poc.py tests/test_p22_file_backed_buffer.py tests/decode_cache
publish:
needs: [tag, tag-tests]
if: needs.tag.outputs.publish == 'true'
runs-on: ubuntu-latest
permissions:
contents: write
steps:
- uses: actions/checkout@v4
with:
ref: v3.9.10
persist-credentials: false
- uses: actions/setup-python@v5
with:
python-version: '3.11'
- name: Build GitHub ZIP and verify source manifests
run: |
python - <<'PY'
import hashlib,pathlib,subprocess,zipfile,tomllib
root=pathlib.Path('.')
assert tomllib.loads((root/'pyproject.toml').read_text())['project']['version']=='3.9.10'
for name in ['MANIFEST.sha256','REGISTRY_MANIFEST.sha256']:
rows=(root/name).read_text().splitlines()
for row in rows:
h,p=row.split(' ',1);assert hashlib.sha256((root/p).read_bytes()).hexdigest()==h,p
print(name,len(rows),'matched')
files=subprocess.check_output(['git','ls-tree','-r','--name-only','HEAD'],text=True).splitlines()
target=root/'ComfyUI-H3-Continuum-v3.9.10.zip'
with zipfile.ZipFile(target,'w',zipfile.ZIP_DEFLATED,compresslevel=9) as z:
for p in files:z.write(root/p,'ComfyUI-H3-Continuum/'+p)
with zipfile.ZipFile(target) as z:
assert z.testzip() is None
assert set(z.namelist())=={'ComfyUI-H3-Continuum/'+p for p in files}
assert all(z.read('ComfyUI-H3-Continuum/'+p)==(root/p).read_bytes() for p in files)
pathlib.Path('SHA256SUMS.txt').write_text(hashlib.sha256(target.read_bytes()).hexdigest()+' '+target.name+'\n')
print('GitHub ZIP exact:',len(files),'files')
PY
- name: Install pinned Registry CLI
run: python -m pip install comfy-cli==1.22.0
- name: Build and verify Registry archive
run: |
comfy --skip-prompt --no-enable-telemetry env
comfy node pack
python - <<'PY'
import hashlib,pathlib,zipfile
root=pathlib.Path('.')
declared={p:h for h,p in (r.split(' ',1) for r in (root/'REGISTRY_MANIFEST.sha256').read_text().splitlines())}
with zipfile.ZipFile('node.zip') as z:
assert z.testzip() is None
assert {i.filename for i in z.infolist() if not i.is_dir()}==set(declared)|{'REGISTRY_MANIFEST.sha256'}
assert all(hashlib.sha256(z.read(p)).hexdigest()==h for p,h in declared.items())
assert z.read('REGISTRY_MANIFEST.sha256')==(root/'REGISTRY_MANIFEST.sha256').read_bytes()
print('Registry ZIP exact:',len(declared),'payload hashes matched')
PY
- name: Publish GitHub Release
env:
GH_TOKEN: ${{ github.token }}
run: gh release create v3.9.10 ComfyUI-H3-Continuum-v3.9.10.zip SHA256SUMS.txt --verify-tag --title 'v3.9.10 — Spectrum reuse and Reference assignment display' --notes-file docs/RELEASE_NOTES_V3910.md --latest
- name: Download and verify public GitHub assets
env:
GH_TOKEN: ${{ github.token }}
run: |
gh release download v3.9.10 --dir public-download --pattern 'ComfyUI-H3-Continuum-v3.9.10.zip' --pattern SHA256SUMS.txt
python - <<'PY'
import pathlib,hashlib
for name in ['ComfyUI-H3-Continuum-v3.9.10.zip','SHA256SUMS.txt']:
assert pathlib.Path(name).read_bytes()==(pathlib.Path('public-download')/name).read_bytes(),name
print('Public GitHub assets match verified local ZIP and checksum')
PY
- name: Publish verified Registry package
env:
REGISTRY_ACCESS_TOKEN: ${{ secrets.REGISTRY_ACCESS_TOKEN }}
run: |
test -n "$REGISTRY_ACCESS_TOKEN"
comfy node publish --token "$REGISTRY_ACCESS_TOKEN"
- uses: actions/upload-artifact@v4
if: always()
with:
name: publication-v3.9.10
path: |
ComfyUI-H3-Continuum-v3.9.10.zip
SHA256SUMS.txt
node.zip
+1 -1
View File
@@ -2,7 +2,7 @@
## Scope
MiniMax H3 Continuum for ComfyUI. Package 3.9.4 provides the V3.9 Sampler and Reference Images helper, while retaining the separate V3.8X2 workflow. Source publication does not imply a GitHub Release or Registry package publication.
MiniMax H3 Continuum for ComfyUI. Package 3.9.10 provides the V3.9 Sampler and Reference Images helper, while retaining the separate V3.8X2 workflow. Source publication does not imply a GitHub Release or Registry package publication.
Sampling Contract v6 and Graph Contract v4 apply to new generation. Keep original v5 history readable, without reusing its Takes in v6 generation or rewriting stored manifests. Public node IDs, socket/widget order, official workflows, Run Storage v3 and State/Session formats remain unchanged.
+61 -9
View File
@@ -1,15 +1,67 @@
# Changelog
## Unreleased
## 3.9.10 — Spectrum reuse and Reference assignment display (2026-10-20)
- Make Second Pass tiled-upscale schedule precedence explicit: a connected
dedicated upscale SIGMAS socket wins, followed by the existing main SIGMAS
socket, while the other tiled controls use their node widgets when their
sockets are not connected. The status report now identifies the sigma source.
- Add the DUMAS H3 Sequence Prompt Builder with clip tabs, local timeline
slots, and one canonical absolute Timeline prompt output.
- Refine its UI around the CSGlide shot-timeline pattern: one multiline shot
editor, coloured shot bands, and shared boundary sliders.
- Normalize CRLF source line endings in both Spectrum file and module-closure audits. Keep the pinned source identities and existing Python 3.13 profile bytes unchanged.
- Add separate audited Python 3.12 profiles for the supported local and upstream 0.2.29 implementations. Keep unknown source/runtime state non-reusable, preserve old unsafe/v5 Takes, and require a new Run for newly supported configurations. Sampling v6, Graph v4, public nodes and workflow bytes are unchanged. No new GPU acceptance is claimed for this repair.
## Unreleased — external Chunks Reference assignment table
- Treat externally connected Chunks as unresolved in the V3.9 Reference Images
helper. Display assignment range 1–16 with the reason and explicit All/Off
range labels; retain direct-count tables and saved future assignments.
- Keep existing selectors, backend routing, Sampling/Run Storage contracts,
public inputs and official workflows unchanged. Loading or refreshing does
not automatically expand old selections.
## 3.9.9 — upstream KJ Preview Override / Review continuation (2026-10-20)
Fix new-Run v6 read-only with the separately audited upstream KJ profile, and
show saved reasons when reuse is unavailable. Keep the original local profile,
Sampling v6 guards, old Takes and official workflow bytes.
CPU: **2,075 passed / 2 skipped / 2 known expected failures**, plus overlapping
90-case JS coverage. Upstream KJ GPU and normal-ComfyUI browser Continue both
passed: **Chunk 1 preserved, Chunk 2 generated, one reused / one new**; all four
AV outputs fully decoded. Browser checks used the existing local KJ profile.
These are targeted repair checks, not general quality or all-provider acceptance.
Restart ComfyUI and hard-refresh after updating. On Preview Override, disconnect
optional VAE/Audio VAE and set Tiny VAE `none`; retain generation VAEs. Correct
unsupported settings before starting a new Run. Old unsafe/v5 Takes remain
readable and are not automatically made reusable. No workflow rewiring or Take
deletion is required. [Full release notes](docs/RELEASE_NOTES_V399.md).
## Unreleased — V3.9 Turbo Sample (2026-10-16)
- Add a separate user-supplied Auto / 2 x 8s Turbo dance and Thai rap sample, gallery copy, ZIP and setup notes. Preserve standard workflows, saved Takes and product code. This source-only sample has no new GPU acceptance and does not replace v3.9.8 release assets.
## 3.9.8 — Issue #31 Preview Override Phase A (2026-10-16)
- Post-release documentation: after installing and restarting published tag `58fbd855`, both PyTorch/KJ and Kitchen + audited local Spectrum/KJ passed new-Run, separate-Queue Chunk 1 to 2 continuation with retained previews, unchanged prefix/contracts and four fully decoded AV outputs. Record the limited audited profile and unverified browser/quality/performance scope. Keep the original tag/assets and existing Takes/workflows.
- Add a finite KJ Model Preview Override signature for the explicitly audited source profile. Preserve fixed preview settings, loaded implementation identity and cross-key/list wrapper execution order; omit only the reviewed builtin `frames` list.
- Limit reuse support to VAE/Audio VAE disconnected, KJ Tiny VAE `none`, and Core NoPreviews/Auto/Latent2RGB. Unknown source, altered helpers, additional state and approximate-VAE preview remain non-reusable. Generation itself retains the existing fallback.
- Keep unaffected MODEL signatures, Sampling v6 guards, old unsafe/v5 history and official Workflow bytes unchanged. Full real-Core CPU gate: 1,995 passed / 3 skipped / 2 existing Inspector expected failures / zero unexpected failures and CUDA initialization attempts; the 48 dedicated KJ tests passed without skips. Normal ComfyUI GPU acceptance passed four Core-API Queues: PyTorch and Comfy Kitchen INT8 + the audited local Spectrum implementation, each with separate-Queue Chunk 1 to 2 continuation and held preview images. Both second Queues reused one chunk and generated one; original Chunk 1 bytes and contracts were unchanged. All four video/audio outputs fully decoded. See docs/KJ_PREVIEW_OVERRIDE_V398.md for exact support and limits.
## 3.9.7 — external-input regeneration cache (2026-10-15)
- Fix automatic Full Run variations when Chunks or Variation Nonce come from external nodes. Unresolved linked inputs no longer let Core skip an explicitly requested regeneration. Earlier accepted chunks, existing Takes, public inputs and official workflows are preserved. Update the custom nodes, restart ComfyUI and refresh the browser; no workflow changes or history deletion are required.
- CPU: 1,948 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and zero CUDA initialization attempts. Targeted Core-API GPU queues verified external Chunks=2 / Nonce=0, repeated identical automatic variations 1 then 2, and 1 reused / 1 generated chunk per variation. Browser save/reload and Run-button submissions were checked separately. GPU scope: PyTorch attention, Spectrum OFF, 512x512, 8-step Turbo, two 5-second chunks. This is a targeted cache repair, not general image-quality or all-accelerator acceptance.
## 3.9.6 — regeneration intent and automatic variation cache (2026-10-15)
- Preserve explicit Regenerate From and Variation Nonce across V3.8/V3.9 workflow restoration and Queue submission. Persist only the selected intent as workflow metadata; keep the legacy unowned Chunk 1 / nonce 0 cleanup and one-shot restart behavior.
- Execute Full Run explicit regeneration with automatic nonce 0 through Run Storage on every Queue. Ordinary Full Run and fixed positive nonces retain Core caching. Sampling v6 safety, accepted Takes, public inputs and official workflow files are unchanged.
- Full CPU gate: 1,908 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and CUDA initialization attempts. Normal ComfyUI browser save/reload retained Chunk 2 / nonce 5 and Chunk 1 / nonce 0 after history loading. Five new-Run GPU queues passed continuation, Chunk 2 regeneration and two identical Full Run / nonce 0 submissions; earlier Chunk 1 bytes were preserved. GPU scope: PyTorch, Spectrum OFF, 512x512, 8-step Turbo; all five media outputs fully decoded. See docs/REGENERATE_INTENT_V396.md.
## 3.9.5 — Issue #31 Attention and Spectrum reuse signatures (2026-10-15)
- Observe the loaded Core PyTorch and Comfy Kitchen INT8 attention overrides with a finite local descriptor, including the verified wrapper self-reference, implementations, defaults, closure, both container dispatch paths, numerical helpers and provider state.
- Keep Sampling v6, runtime signature v1 and Graph v4 unchanged. Overrides outside this adapter retain their existing observation policy; unrelated incomplete MODEL/CFG observations still disable reuse.
- Add a dedicated Spectrum binding identity for explicit audited provider implementations, including upstream Spectrum 0.2.29. Include every effective immutable config value and verified loaded functions/wrappers; exclude reviewed execution-local history and counters. Unknown provider sources, extra runtime state or altered callables remain non-reusable.
- Preserve original unsafe Takes and v5 history. Only new correctly identified runs can gain reusable prefixes. Public nodes, workflow bytes and storage schemas are unchanged. See the versioned validation record for CPU/GPU coverage and remaining limits.
## 3.9.4 — Registry packaging repair (2026-10-14)
+13 -8
View File
@@ -1,10 +1,15 @@
# DUMAS H3 Continuum Tiled fork notice
# Repository notice
This repository is a personal fork of
`ukr8b3g-cmyk/ComfyUI-H3-Continuum`, retaining its upstream license and
attribution. The fork adds the `H3 Continuum Tiled Latent Upscale (Per Clip)`
node and changes package metadata to identify the Dumas repository.
This repository is now an upstream-compatible mirror of
`ukr8b3g-cmyk/ComfyUI-H3-Continuum`.
The added node is intentionally an adapter around ComfyUI's installed native
`MMH3UltimateUpscale` node. It does not redistribute model weights or copy the
upstream model implementation.
The former DUMAS fork experiments remain available in git history, but are no
longer installed from this branch. Install the maintained upstream repository
and the separate DUMAS custom-node repository when you need the additional
MiniMax H3 nodes:
- Upstream: `https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum`
- DUMAS nodes: `https://git.dumas.ddns.net/chris.dumas/DumasNodes`
DUMAS H3 nodes are grouped under `Dumas/MiniMax H3/...` and use unique node
IDs, so they do not replace or shadow upstream Continuum nodes.
+15 -3
View File
@@ -1,16 +1,28 @@
ComfyUI-H3-Continuum 3.9.4 — V3.9 / V3.8X2 インストール
ComfyUI-H3-Continuum 3.9.9 — public KJ / Review continuation
CPU 2075 passed / 2 skipped / 2 known xfailed; upstream KJ GPU and installed local-KJ browser Continue PASS.
既存Take・Workflowは維持。更新後は再起動・ブラウザー再読込。旧unsafe/v5は自動変換しません。
対応条件と検証範囲: docs/RELEASE_NOTES_V399.md / docs/KJ_PREVIEW_OVERRIDE_V398.md
**最新版パッケージは3.9.4です。** 固定版のGitHub配布は下表を参照してください。Registryの公開状態は別に確認します。
--- Historical package record (unchanged) ---
3.9.8: 監査済みKJ Model Preview Overrideの画像保持後のChunk継続を修正。対応条件・検証範囲:docs/KJ_PREVIEW_OVERRIDE_V398.md
ComfyUI-H3-Continuum 3.9.8 — V3.9 / V3.8X2 インストール
**最新版パッケージは3.9.8です。** 固定版のGitHub配布は下表を参照してください。Registryの公開状態は別に確認します。
| バージョン | Release | カスタムノード本体ZIP | SHA-256チェックサム |
| --- | --- | --- | --- |
| 3.9.8 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.8) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/ComfyUI-H3-Continuum-v3.9.8.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/SHA256SUMS.txt) |
| 3.9.7 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.7) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/ComfyUI-H3-Continuum-v3.9.7.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/SHA256SUMS.txt) |
| 3.9.6 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.6) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/ComfyUI-H3-Continuum-v3.9.6.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/SHA256SUMS.txt) |
| 3.9.5 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.5) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/ComfyUI-H3-Continuum-v3.9.5.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/SHA256SUMS.txt) |
| 3.9.4 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.4) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/ComfyUI-H3-Continuum-v3.9.4.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/SHA256SUMS.txt) |
| 3.9.3 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.3) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/ComfyUI-H3-Continuum-v3.9.3.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/SHA256SUMS.txt) |
| 3.9.2 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.2) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/ComfyUI-H3-Continuum-v3.9.2.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/SHA256SUMS.txt) |
元の[3.9.2 Source ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/archive/refs/tags/v3.9.2.zip)は従来のタグから取得できます。本体ZIPと公式Workflow ZIPは別の配布物です。
3.9.4はRegistry配布の梱包修正版です。ManagerでH3 Continuumを検索し、3.9.4を選択するか、comfy-cliで `comfy node install comfyui-h3-continuum@3.9.4` を使用できます。Registryの検査状態とManagerの一覧反映は別に確認してください。生成・保存動作は3.9.3と共通です。
3.9.8は監査済みKJ Preview Overrideの画像保持後の継続を修正します。VAE/Audio VAE未接続・Tiny VAE noneが条件で、KJNodes全般への対応ではありません。更新後は新Runを作成してください。ManagerでH3 Continuumを検索し、Registryの検査・一覧反映を確認して3.9.8を選択するか、`comfy node install comfyui-h3-continuum@3.9.8`を使用します。更新後はComfyUIを再起動し、ブラウザーも再読込してください。既存Run/Takeの削除は不要です。旧markerなしChunk 1/Nonce 0は選び直して保存してください。旧unsafe履歴は引き続き再利用不可です。
3.9.3は一時ストレージの保存先設定と状態表示・未使用削除を追加した版です。V3.9ノードと3.9.1/3.9.2のWorkflowはそのまま使用できます。旧v5 Takeと履歴は読めますが、新しいv6生成の前半には再利用できません。新規Full VideoまたはStart again from Chunk 1で開始してください。新しいGitHub Release/Registry公開は別工程で、過去のRelease/tagは残します。V3.8X2 WorkflowをV3.9へ自動変換せず、別々のWorkflowとして保存してください。保存済みRun/Takeも自動移行しません。
+133 -98
View File
@@ -1,10 +1,23 @@
02ea0951ce2e7ac2d5f3bbf02c596a9f9ea8adcdf2a113dca4c8db3cc19c36b3 __init__.py
48c25e689d401e4764fea0084100cdd43f8ff524410d10a9be49d3befb8b7202 .comfyignore
ce319f06573b0b53709bae56d9b9ee0b3f4180eba961d7cbbac9e01e403a589e .comfyignore
3d681ded5d4c50cc20c2382e29fdc98d8a48e76d552c41ad5a195d941e66e612 .gitattributes
c3fc30c17dc92c4c772f25392f3f63385ccd5de025798d13dcea51c03f1f81e8 .github/workflows/publish_registry.yml
60392b0310924b7abcffbfa8070f0180e531a1fb1c030d85496a62b46291064e .github/workflows/publish_release.yml
3e22b441fe96bb6d099f0fcb92265f7a986c444f6c45be263c7905898aea4ac7 .github/workflows/tests.yml
a28c4c592d56c57e0967f0adce5655313b942b8cc04f77766c0db1bec96c13e0 .gitignore
fa9c99303d1a960db12ae2191cffcfbffe82e25e144eb6d16775c840fec4d573 CHANGELOG.md
6e889c42490ccb5f376694b5944c693a84e7febedf38e631d687e96bc1e76376 H3-Continuum-LLM-System-Prompt-v1.zip
144324850d4ba086938da5f4ed7b02ff0668dec554f092695328240e43fec3ec H3-Continuum-Skill-v1.zip
fe29dafce4a6c5229e7e3aef1f946afba2346015702ed8cdda9927a80b6e65f3 INSTALL_JA.txt
c44abbd789091bbead01250dbd5453e60539d2665c69f162335fdbd2936019c2 LICENSE
656e417909695c5dcabf134b6f3240c44bd02bb2e33881b5247efd345cb932c2 PACKAGE_INFO_JA.txt
57fa9ad77ab5b1d2da6ec43f7ba19118e253cb4a959b758b246755471df15591 PACKAGE_VALIDATION.txt
da45eeb9f4950777bb63974cbf1d77856b287b05e3466be63571f2ad4561f12e README.md
eed26374c7de5e9e19396836e0270fb46f37f3302dfd6f5533c8d703741cbf0f README_JA.md
9a84b03b89f5889f2ed59bcb969711bc1de367ac109b14596483283d3ffa1174 REGISTRY_MANIFEST.sha256
0dbe8a5f71c92f3523942343b40787104c8b96781461ba3ba82934a6bdbe5834 THIRD_PARTY_NOTICES.md
02ea0951ce2e7ac2d5f3bbf02c596a9f9ea8adcdf2a113dca4c8db3cc19c36b3 __init__.py
7fd122ff08232e114a6ef32171217ea6324e980eb7257d560c466b8177a3d3ec audio_compat.py
3e7a1ec3380be2a222c8a3e9b8a6bb184d3a760478957f48186438f15235bdba branch_provenance.py
1358bcdd91ba8c4ee29c92c203bcdea9c34fd0ad7437bc0f00b657ce0ae32c72 CHANGELOG.md
c1ed3a0342730450958fb62fd21854a876991f3a4aefb24e5383b0c73cd2312d compatibility.py
4d62508f900af9f93ec320afb900c27cc9a300d1e0c0a94714258498c5ede5fc conditioning.py
294251c40b1be2e296c5e5f627b68091d087bdb39179e3a0c3cebd2ee8a344f5 constants.py
@@ -19,6 +32,39 @@ c93a3e52e4deabe5ef0db7f34a7605856a543089574ba17227e697e9a2c0139a docs/ARCHITECT
1d60cf5d4ef9c45885d4f90289c3b17aa70a6a2752c6b288288322a5e0568050 docs/FAILURE_SEMANTICS_BASELINE.md
c6ea788a03c5f29f32fb6a4fd62db92f108e604493d818b9729e9d7fb967526e docs/FAILURE_SEMANTICS_DELTA_R5.md
7660570840229b6fcbe62f0fbb52966d0099d64d7e6f01a739fbf9899293c8e4 docs/FUTURE_ROADMAP.md
44377cbac9131daa3d1c88ccc154c325257767f4c52f3181f61525c7eae00531 docs/ISSUE31_SAMPLING_SIGNATURES.md
b7ecd069d803cae7f86ad0ab89df144b5a5b157c168c73ee2962b47234d4ec0c docs/KJ_PREVIEW_OVERRIDE_V398.md
77c225d1e49e06d00b04acf811177c9a4317e792d4bfe92df539de99420c3563 docs/LOADER_PERSISTENCE_REPAIR.md
b30fe4640359bebfcdeeb94dbeec94cb346c8e3646509e1dcdb87d7021ba04ff docs/N2B_BRANCH_PROVENANCE_CONTRACT_V1.md
994b0b969089247921ca21306496cc1e8e6f729b587a8206b6403e0b7a11734c docs/R0_RUNTIME_CHARACTERIZATION.md
e6b8cc3250776868f473357cd97ce299fcea31ae18b879ec6e6466f47dbf704e docs/R6_NODE_FACADE_DELTA.md
bce0337cce903938b6b8df7f52c5c67e2d0ee14599c693687e4180debc606db2 docs/README_HISTORY.md
9ee92cdd3783e61e3bc71b99d5c0df7dd0a9f129c8a6b37701d7001a4f745e77 docs/README_HISTORY_JA.md
cebf8462681796df807987d8afef2c78abb0953b54b2f5dcd12d583bde0ec806 docs/REGENERATE_INTENT_V396.md
bbe74f802339b758d8874900e2617dd1a46844a58e0f872d95f34b9b40ba20ed docs/RELEASE_NOTES_V3910.md
1e3156127e570f7eeda696cf1143f42a4b3144f3ab3a32c6e2deee2ed9e8e3be docs/RELEASE_NOTES_V395.md
6de7daf79d0cf8e8bf03d01d030bc0db43df454b644c7fb277c4caa53bf0dae4 docs/RELEASE_NOTES_V396.md
8b6f43df28d660e44e6d4bdecd161157091e8f30bd2305c686918d288c9a9f1e docs/RELEASE_NOTES_V397.md
2e1885630db223593acf95c4e6ba0d6db55cf5806601fab94fe3b146307aa89f docs/RELEASE_NOTES_V398.md
49d02b7650d4a542307c9d4189dc217ba1db54c4eac959cc2e2a1ce57b501ae3 docs/RELEASE_NOTES_V399.md
91ee4adf828415e84e2f413e3a4e20dc16356f139e8ef1f6756f584cad04f3c8 docs/REVIEW_RESTORATION_REPAIR.md
93f07aa727b34802bce3d333426b8ba2af522500882a6d222d00cf577ca82e49 docs/SEAM_TEST_SOURCE_JA.md
4546ffb0cbaf270c52d1900908e46bfa2d3c0e2d0035ae5694680890e89aedd8 docs/TEMPORARY_STORAGE_JA.md
4c33d223a77c78e5c74946aa43f537031fd5aa415388d46fed975f9b07d53c3e docs/TIMELINE_VIDEO_EXPERIMENTAL.md
00f88cb57be5e79d9c881671bd13fa154d91843f4c1ae2f2451ee2a73841271d docs/TURBO_SAMPLE.md
92bee3f3c0c340356abdd06b4993fbf2e98cbe57e931f85cbfd11da0f7f87809 docs/USER_GUIDE.md
c3f0753bdae1c54044a6e9ad5770c45e85a306ac9c15cd363790f8625c865f94 docs/USER_GUIDE_JA.md
2c13aefd137db92a3dbf5576e9442d58012b1f0de1240c07d039ceeb702e7c1f docs/V2_1_4_UI_SIMPLIFICATION.md
b08022df068576c759c8bf80ac6b6519693f3c65c917f1822aeb06ea359aa5b1 docs/V2_1_5_UI_SERIALIZATION_STABILITY.md
a032738b0bc5acb136e92d8bf7694643a5f518b9c7fc0fc878d66d81ba8b1b04 docs/V2_1_6_UNIFIED_UI.md
384e3841603f7c6c8286ef147e93aaf7d056a286503a2a4fc21f397e4dec4876 docs/V34_DRIVING_AUDIO_VALIDATION.md
1840c81b4090ff3c86c0c1b0acc2cc10d3d9839e383c81f262a5c59a67504cd6 docs/V35_HIRES_FIX.md
ca9f1aecd1032f0f0b237948c7fc1e4e693440dff35f5f17fd5e6d5ca4dab7f9 docs/V38_BEGINNER_MANUAL_PLAN_AND_CAPTURE_SCRIPT_JA.md
717577efdec5e61600504b77bc2a099ef022c4fe2333e936c3310c7182d8f2a0 docs/V38_OPEN_INTEGRATION_CONTRACT.md
3496636f645b39336fd95a4f21a562810c56f8e1d92fa6eeb1e4b0786ad11b37 docs/V38_RELEASE_AND_MIGRATION.md
4d17d74107e78404b190fec9494b443249e5aff4f58670c50d54dfdfb064334b docs/V39_REFERENCE_INLINE.md
79bf8886f91eb3e8e7e8db8a8093eaa12fbb012b156e8d478de14f089f7c19f4 docs/VALIDATION_JA.md
3fec541a675c1e4e5e287cc8e498036001f34f7a62460321e16f9ba27b5e4c18 docs/VALIDATION_RESULTS.md
9a634b4ff6c9ca98285f6b250402a3d0e67d29f031e134d9bfcab54c119ed337 docs/images/comfyui-manager-h3-continuum.png
6c72d9a9e35648355c4dd4c8f060a6b7667e322f4b6c86690df9d6769ba182b7 docs/images/h3-continuum-advanced.png
b9f1dca35706a1cd963ba892d218568ea30173c6cf6a3bdc896f7a2266eb9407 docs/images/h3-continuum-detailed-report.png
@@ -73,92 +119,88 @@ a4705ae4b5730d2b43e29fe125a64196a031f14bcfa8702663a17aa89f683d89 docs/images/v3
ae3338f720fc22c05ab4a661c68f9c5dcb20d751a835f0e60d4738333be48e63 docs/images/v39/v39-feature-summary.png
84304d318eea6db3921d366647c6fff9ad488e8c4cfef6ca00988bfd2a6504bc docs/images/v39/v39-reference-images-node.png
b765b37502b012b462606c4d2dcef43f13950e5fd682824fb18cdb19d542c121 docs/images/v39/v39-sampler-node.png
77c225d1e49e06d00b04acf811177c9a4317e792d4bfe92df539de99420c3563 docs/LOADER_PERSISTENCE_REPAIR.md
b30fe4640359bebfcdeeb94dbeec94cb346c8e3646509e1dcdb87d7021ba04ff docs/N2B_BRANCH_PROVENANCE_CONTRACT_V1.md
994b0b969089247921ca21306496cc1e8e6f729b587a8206b6403e0b7a11734c docs/R0_RUNTIME_CHARACTERIZATION.md
e6b8cc3250776868f473357cd97ce299fcea31ae18b879ec6e6466f47dbf704e docs/R6_NODE_FACADE_DELTA.md
91ee4adf828415e84e2f413e3a4e20dc16356f139e8ef1f6756f584cad04f3c8 docs/REVIEW_RESTORATION_REPAIR.md
93f07aa727b34802bce3d333426b8ba2af522500882a6d222d00cf577ca82e49 docs/SEAM_TEST_SOURCE_JA.md
4c33d223a77c78e5c74946aa43f537031fd5aa415388d46fed975f9b07d53c3e docs/TIMELINE_VIDEO_EXPERIMENTAL.md
2c13aefd137db92a3dbf5576e9442d58012b1f0de1240c07d039ceeb702e7c1f docs/V2_1_4_UI_SIMPLIFICATION.md
b08022df068576c759c8bf80ac6b6519693f3c65c917f1822aeb06ea359aa5b1 docs/V2_1_5_UI_SERIALIZATION_STABILITY.md
a032738b0bc5acb136e92d8bf7694643a5f518b9c7fc0fc878d66d81ba8b1b04 docs/V2_1_6_UNIFIED_UI.md
384e3841603f7c6c8286ef147e93aaf7d056a286503a2a4fc21f397e4dec4876 docs/V34_DRIVING_AUDIO_VALIDATION.md
1840c81b4090ff3c86c0c1b0acc2cc10d3d9839e383c81f262a5c59a67504cd6 docs/V35_HIRES_FIX.md
ca9f1aecd1032f0f0b237948c7fc1e4e693440dff35f5f17fd5e6d5ca4dab7f9 docs/V38_BEGINNER_MANUAL_PLAN_AND_CAPTURE_SCRIPT_JA.md
717577efdec5e61600504b77bc2a099ef022c4fe2333e936c3310c7182d8f2a0 docs/V38_OPEN_INTEGRATION_CONTRACT.md
3496636f645b39336fd95a4f21a562810c56f8e1d92fa6eeb1e4b0786ad11b37 docs/V38_RELEASE_AND_MIGRATION.md
4d17d74107e78404b190fec9494b443249e5aff4f58670c50d54dfdfb064334b docs/V39_REFERENCE_INLINE.md
79bf8886f91eb3e8e7e8db8a8093eaa12fbb012b156e8d478de14f089f7c19f4 docs/VALIDATION_JA.md
3fec541a675c1e4e5e287cc8e498036001f34f7a62460321e16f9ba27b5e4c18 docs/VALIDATION_RESULTS.md
a8c56379abe23b98c8ceed6ea5e60a02ecc3c6d9636232d7371615f4eb306e81 driving_audio.py
6771c9905d35666121c4892d2ec26202505f96d244ce814ad9a87a02878952cd examples/H3_CONTINUUM_V1_WIRING.md
c3cefcdf50df9d6609f583a008453b587c011a0fb41e290d5f1196ce62f58877 examples/H3_CONTINUUM_V2_WIRING_JA.md
be419d6c773784823af7b58a3423ef2e21d3cc3e23c40e60e274560895ca034b examples/H3_Continuum_V2_README.txt
4b94f779a98bfc555f6999113a694da07746cc3cf400bfc70ebc580592034d52 examples/H3_Continuum_V2_reference.webp
c3cefcdf50df9d6609f583a008453b587c011a0fb41e290d5f1196ce62f58877 examples/H3_CONTINUUM_V2_WIRING_JA.md
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/MiniMax_H3_Continuum_V38X2.json
40199d2500174ba51c040d6983f88d74c4bfbaf19191e728818e6b969b869b0f examples/MiniMax_H3_Continuum_V39.json
705d1141842863b82abe26a8bb015d45eff3bc81ca5bdc53e58b464b743173c0 examples/MiniMax_H3_Continuum_V39_Turbo_Sample.json
68e1a724d3b99458b86b8f643dcb916183e7e8dd12d7b4b3850991feb8077694 examples/settings/recommended_settings_v1.json
5a26408c770381cdb259cb710de7dab12d7ddecca262bdbf35ba1566056050f5 examples/settings/recommended_settings_v2.json
b8275c8cecde12aee0418cea456d1e7f4b875cbb1bce32f1f3d9e06a44dd0be1 examples/v2_prompt_list_example.txt
0e59ea821081cb93165aa3f88c6949736dd225837fe1cdcf1173aa5cd54c6d29 examples/v2_prompt_timeline_example.txt
26bbca92b5f7cac4cbde89290b3f801b7a895df76a07493ca6f65c18d5b1154a examples/workflows/MiniMax_H3_Continuum_V38x.json
df4e0e355020d2e731932cff86a4b00932b4f193ec7ed513be0db994ec0e71bb examples/workflows/MiniMax_H3_Continuum_V38x.zip
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json
4fa8680037595f55f97715f9a2f63d28f11328a87f5c33c316cc4b0ae817e668 examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2.json
f3d8dd0f0872afdd97fe24e58b5cb74190d03e3583a66a5d013fee41f6894cda examples/workflows/MiniMax_H3_Continuum_V38X2.zip
171f0b5596724926d1d89fd12a46434f6010f12e651d19865b9b6160a1e9fff6 examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Follow.json
e61e45f218609a7ec845d2e8762dfd1efed6c51a398061066d1e51fc93556663 examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Follow.zip
7d6754cfc75599e07046b1949c1723a56f57871ecd5abc9543fcf8b34bca8c04 examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Repeat.json
9b60c8495e3d0e32c2bcd609080342ae19d0fd4c7c0662d5fe32e7fad4a9b41c examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Repeat.zip
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2.json
f3d8dd0f0872afdd97fe24e58b5cb74190d03e3583a66a5d013fee41f6894cda examples/workflows/MiniMax_H3_Continuum_V38X2.zip
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json
4fa8680037595f55f97715f9a2f63d28f11328a87f5c33c316cc4b0ae817e668 examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip
67e81fbdd4c21a8d64fb3bedb5e495698fc8df5cdd9540c8ef5476caa76119c0 examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.json
65689fd800a60ab6939bea020d630616b03a453cf6d48022cd1361992a765abc examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.zip
26bbca92b5f7cac4cbde89290b3f801b7a895df76a07493ca6f65c18d5b1154a examples/workflows/MiniMax_H3_Continuum_V38x.json
df4e0e355020d2e731932cff86a4b00932b4f193ec7ed513be0db994ec0e71bb examples/workflows/MiniMax_H3_Continuum_V38x.zip
40199d2500174ba51c040d6983f88d74c4bfbaf19191e728818e6b969b869b0f examples/workflows/MiniMax_H3_Continuum_V39.json
f69b734b5bf19fb6db78bbe4737db0bef69c2a6cef2710f28767882ef35ec847 examples/workflows/MiniMax_H3_Continuum_V39.zip
67e81fbdd4c21a8d64fb3bedb5e495698fc8df5cdd9540c8ef5476caa76119c0 examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.json
65689fd800a60ab6939bea020d630616b03a453cf6d48022cd1361992a765abc examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.zip
705d1141842863b82abe26a8bb015d45eff3bc81ca5bdc53e58b464b743173c0 examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.json
2096e583c6805c0f6f54f76a81e0ed1246325a16d189afba0423fa05e2c19599 examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.zip
a344da581b3c2d77558a460d2b5256e6c9a63b5008c98b52999ceda2e0b2b767 graph_contract.py
bf0556b076dbaca67c3405a7d86c333ca7fdf3d4381529b0a9edc906519943e9 guide_timeline.py
6e889c42490ccb5f376694b5944c693a84e7febedf38e631d687e96bc1e76376 H3-Continuum-LLM-System-Prompt-v1.zip
144324850d4ba086938da5f4ed7b02ff0668dec554f092695328240e43fec3ec H3-Continuum-Skill-v1.zip
1473c384082700f422f2cf9141e0d012d87d2b43ac8a8a433edc7172e6ca7bf6 hardening.py
33bfd17df1c51c5730ee8007ca3a52e0221d5424b2e86398c50003b9a8f4102c INSTALL_JA.txt
23bddc5a41d2e997f4fbf789c7128e32f4f197f1405b2aef1c76794f57482446 install_windows.bat
12bf572a65d57bbc9a72b459f29691c2057768f4fb7a74f9db87f4775e995bda install_windows.ps1
139546ec1ff459945aa89b50f74b5ee69f62a0e07bbddfcae06936091a409a83 layout_adapter.py
c44abbd789091bbead01250dbd5453e60539d2665c69f162335fdbd2936019c2 LICENSE
d3e213005670e94dee33a240349a4d245cf9a3b6602099f11ed30cf5a32e2c21 media.py
52cec83cfffa8d2504c1f64a9614fb59e56b62c4b1f3106fd3b4fa73870ba6e1 metadata.ini
7ad8ea8859b666cb2771f19795ca8c04ab7aafb5c20cff4e80ad17295429ded3 metadata.ini
fd23c7f32810788aaaf1709c4959c40ec72303aecc6f6afa5fa68d824cca97a8 minimax-h3-continuum-dance-director.zip
d7e613e2cc24110213b477bdd1d390234a41f2af3b1ed006a809066d71d658eb model_patch.py
ecb65a5dbf9e1683c5a3cadd82daaea3a90a2b776f6ebabfbe8b07aa93177422 nodes.py
4decb6e44ee6d172352f223abca4c7a8cba2c8ac6ab0f325b8645f60609f60f1 PACKAGE_INFO_JA.txt
069bf25bd44e7b119c285740adf698bbf30d722a9ca636cd78fe06f026b07d67 PACKAGE_VALIDATION.txt
6ba95639b52219fc37b4f51d52cf6879f75c7418a8d8815b3725b3f201a4f1e6 preview_override_provider_public_v1.json
d4cf474cbd1a98867075c45b53b2ab4d47752bb11611c916bac2a6831ca82c0b preview_override_provider_v1.json
3b7e18bd2ba50fd6c8fe71c58ca79a0bf752f362ed78b7357ce1998efc934b4b preview_override_signature.py
b470819f16fc179654dc1600721a63f0a5a14e777f4c004d574b499da6886ba5 public_api.py
7ff970c9f50192030a302e4967f1747b373dcf5584e823883ec2cc5f8c1b46a2 pyproject.toml
dc56e02aa94561076bee8a836c268c0e45a8b5248f3b40108bcb740252965ec6 pyproject.toml
94fdc2d545b2d4077a4d70229fd22ba3ccac21d49d1097d2873f76d04e9a3c7a pytest.ini
4adbf81a12fa7c99dc748b4275aeefaa098596bee67eca2d118b15d034fcc211 README_JA.md
6b1732f58c2eb300cd1493b7ecdc450ce23910676793d834453abc1aba39bf7c README.md
caa572ea65dafc86850db99f52433d360ac9c4ab36b854292a3dccd5ba32dc76 reference.py
9539f5cb9fb695841073fa58d9246d5a7410bc372f6cd340378185004dfaf978 reference_audio.py
5cb9bfd6810d3a93a637451980a4cc5c1436e5b84cfd1c7a111cd192e876351b reference_video.py
caa572ea65dafc86850db99f52433d360ac9c4ab36b854292a3dccd5ba32dc76 reference.py
b47df4eb7084237aed3cee27d00704f8349035c6659943adf7cbcc7b86d62832 REGISTRY_MANIFEST.sha256
467edb2c188d024c2392aafe7423b3f835e16b10f570da76e2f7dda8a962c1c5 requirements.txt
1845cbf7ba4d3b35afd087a15f5a4d92dc73d4858e61982f8ccf431a836a22b1 run_storage.py
6056de6c0a5322a8a51ba5a114538dca2ef67788cd1afba10fedd483b95f6afa state_io.py
985e7647bd2626af5df68ca702cf6ebb38cc09f3ffe0e239d3e8ef256352a5d8 run_storage.py
8f786989af6339be8cec8f2bf602da89344cf96202d29c980af5c07175eea3d3 spectrum_provider_v029.json
8cdf37b165eb68462dd7accf4a1fd63043f02479f9c377778fb9d4b6feaddcc6 spectrum_provider_v029_py312.json
0efc8d65cb6f59042457ad843fe1b77a4bee45e59d3732f5c83fbc75a31ae77b spectrum_provider_v1.json
71801234f6d2e08b660803194653821ca648e5b9a0d0c3f138bc7ccc6fe9d8c6 spectrum_provider_v1_py312.json
448213ae7670a6e231f833ca1901ca295eb5f6bfb8983ca37703bdb3fdf15f99 spectrum_signature.py
3a359a0467725cc292ccb4c86cf0c3d30bb6a114d01f5345792b09ea3d45441d state.py
6056de6c0a5322a8a51ba5a114538dca2ef67788cd1afba10fedd483b95f6afa state_io.py
660a9df28c7d62ad383cda5514c48c7efcb3576f435af023040116d533a56563 storage_api.py
517662b104ab1507bb927a39f25b8503f092c2ba4c39bc277379ba9d9e31b334 storage_fs.py
daeda14d9f5338835661e6738e074a3897d936cd847fe3e78519efabbcdfd0ef storage_memory.py
53da3ac42f4a66e8adc4560049c7c8e7ada82599be8c61f0573c234342ca36ed storage_settings.py
43ec4023dc06c941e5d7a04db4b8a89fb58d744ceb986f6eb9bf534d4fdcb87b temporal.py
8a50f5ad8a7f100eb029d74999999270eb52fbffa1be43625150c3b11bf1a72c tests/conftest.py
b2435d87944c3ba8669c81d729e670f1ae78a210114ed864d2acca860529787a tests/conftest.py
7f34c43ea74c43cfc4f9453848c69c8ed7546de5b78c7c7d063d643f2c483c36 tests/decode_cache/conftest.py
8d0cced8c945c4333ef78688b51a988d7e1d560e5f9e6cceec8310b17c98ad53 tests/decode_cache/helpers/cache_process_worker.py
a41a4de12e6fe5690845f9f0795fa82c4907f0afe2551034fa2a4c95617ac39c tests/decode_cache/test_adapter_nodes.py
19d47f94ed2bc56d79b4345f820de3cbb3191e47a5034363a141dd7af500408e tests/decode_cache/test_identity.py
549698d35918b611adc4e8f3aa2f0568c19291b4825cc3c593f3f63e77ec5844 tests/decode_cache/test_process_isolation.py
7e627131904dc6d6f58021fad396185c531858b62dd6febf12c3f95fd1391138 tests/decode_cache/test_store.py
38c2dbf1481d6db0afe8554a4b7c9a27c774ce8fbffcc1fcd001d35709695159 tests/fixtures/issue31_local_spectrum_signature.py.txt
cd9f3f070ab8d4486aecb28c9cf51691b02d4dad4ebf0ecb872469b180eccfa7 tests/fixtures/issue31_phase1_model_options.py.txt
db08ce77cc5c5835f2403d4d589394090fa5d25ff3b1840312b2808e1bc203b7 tests/fixtures/v38_a3c_selective_refine_saved_workflow.json
82efab13baacb9a229671eda5f4a2d24bb839b160b2bfb40591df51a5ddf80a4 tests/fixtures/v39_sampling_v5/project.json
8b8ca1b0c802d024c6915d903747d8722bea25628369997ab008ff896462b96a tests/fixtures/v39_sampling_v5/revisions/d8352287737286e5/chunks/rev-d8352287737286e5-txn-008f52b93a88461481ead25f91627af0-chunk-0002.safetensors
26e771cffc57f171f4cc4c6859876350bdf717d1e8213fc005343006a34f002f tests/fixtures/v39_sampling_v5/revisions/d8352287737286e5/chunks/rev-d8352287737286e5-txn-9bb0a584b7344342b114bd31b175225c-chunk-0001.safetensors
7bf0cf46f049e072a68e71cf8bbab10aef695f0931b16842a890c084021f5a11 tests/fixtures/v39_sampling_v5/revisions/d8352287737286e5/manifest.json
606b3985f2b644fac40c84de3e08d1453e87a988acca51ce9595516a7c60a577 tests/frontend_decode_cache_helper.mjs
868c613654467c2c739e07de5dd9b4679b36f98cd96d631d91e1d4afe5698cf5 tests/frontend_external_int_size.cjs
a328953818ef9bf43e411331d2289ccf98a7bdd31dd9e1c1af449b4a1b632c9d tests/frontend_reference_images_v39.cjs
751f38be4351d459145beb3a533530e0c667552fe148de2f732efbb04494b660 tests/frontend_review_queue.cjs
8d0ccf5711e1be0bf63b4e02fecb0b37cd88979894d27b89337cda4a6d1c4ec5 tests/frontend_issue30_ui.cjs
d84bc6c8f119de1c06257d1c85f135fbea74dfa075b3b9ac8d86daefcf332490 tests/frontend_reference_images_v39.cjs
c4563a9002cc17310226a27790e0ec8c1770b6525b899f398d44fe93e2b56fc6 tests/frontend_review_queue.cjs
ea47f51abe07e9b2af840272939c0f363369636af30f4c6dffb0bb696b752c8a tests/frontend_storage_settings.mjs
36f74c03a40c4b62b90eede070d976bcd1b7f0ea828983d1e61a6cb6d80a0ea8 tests/helpers/r5_process_crash_worker.py
706029af8c8d979b2b700f5bc2b96a3064fe7c728f487698d9f079abb02e9409 tests/terminal_merge_diagnostics.py
02e6286857353b76080b73928c0d1e7ec48b1ed16339231a4a450743f193315b tests/test_audit_boundary_repairs.py
@@ -170,9 +212,13 @@ a328953818ef9bf43e411331d2289ccf98a7bdd31dd9e1c1af449b4a1b632c9d tests/frontend
bfcfd77e70711830d087878d78021b754cd7e1ea324d50af0caa72f041b298fc tests/test_easy_load_video.py
72c7a2ce36d8589f1de386805dba9ef1c12a3f99f9cf818e4752d5d26b39d428 tests/test_external_int_size.py
326a97499ffe51935dd3259dbab4ed71fb02955cd3ddb5a52b5c6558a7efc872 tests/test_gpu_diagnostic_runner.py
4926706cb171e4e36e8d7ed46b592b0ed6c9c2e100072feeba497b651e52653b tests/test_issue13_r2_tools.py
aaed0385281e1782edc96f1d084fa84bc4e7ec9e0055ec35aaf9a9f8b6e5bfce tests/test_issue13_r25_tools.py
ff2e865fb49f0c9f73f3f82eff78ff15373ce8e843514d610b73231dd6217a93 tests/test_issue13_r28_tools.py
4926706cb171e4e36e8d7ed46b592b0ed6c9c2e100072feeba497b651e52653b tests/test_issue13_r2_tools.py
d6b5b969a0c02f1b282da990db3e503e0768747cfac8a4edd32cf8677550e024 tests/test_issue30_frontend.py
6b75fc849ded1b155547b7c371f99f8928bfb180044aa0e98f46e64b1d14d805 tests/test_issue31_attention_signature.py
25683f1e109604807520b85d778947e435aefe68805ce5a1d67c30a3e51dabd8 tests/test_issue31_preview_override_signature.py
0e54b7656568ad01d06251e36b793ce912210b94c7e02f183584461f58a48de1 tests/test_issue31_spectrum_signature.py
c058f189d8e58fd48a80c20c0a857969737a8b13960eee60e5c17db2b9288bc1 tests/test_layout_adapter.py
a9da2d25e9fc9b87b249274d7bf2ab061d14f5d1e9c482b69d795ba511cdace1 tests/test_loader_store_roundtrip.py
ab1ea95e363ad0effb9459bcbe713838f12e3501515871203672ea91e7fd4c32 tests/test_m3_conditioning_display.py
@@ -191,9 +237,11 @@ c2560cbafc5962bce5f063e262cf6a173dd8df18f7b58035ec326f616617d61c tests/test_p25
701cb4c8b43719fac054536fc110b82dcb46939d2135d0a99313993abfc5635d tests/test_p32_profile_server_static.py
403a0ff300330ac1f97642bf0681e80b83eec18e8911c03952ddf203ce794e5a tests/test_p46a_reference_video_memory.py
2d16f6de7bcaf1f92718d3719654a746438f6218bba8d25b3632c318a1ab9c98 tests/test_packed_layout_preflight.py
cd53a068ce0df4aa4c242d5ada4609beac0d90d0b9633ada766607ea684c08fc tests/test_preview_reuse_diagnostics.py
3b9f938de680bc8abd61d1373472589df96112c50f2d46eb0f47e26c46c9f49f tests/test_preview_reuse_frontend.py
1ae0369943ba2fb13f21ad0cf56ee58621ba09bf20174379595e157a1775c7fd tests/test_public_api.py
892d3e3e022a527652dd995508c8bbb650c370caf9ee66fa3eb1f7fd288bba38 tests/test_public_workflow_integrity.py
c9ee25a2f188b59f949c27b95a29a467d583a2dfe92d15f17e9961b6ef696380 tests/test_release_metadata.py
988d4f4973410d387ef66f11619021f8548b0447759ae0dae3f559ed5c65f11f tests/test_release_metadata.py
5f6caf1814052798f20571ebb863c098bec23d8f2e091fc1f30ce1ef9c58675a tests/test_review_identity_lifecycle.py
5f7d3e01bc78fa076ecb17f8532b6d3f3137545da4028032ad6769407d80be07 tests/test_review_provenance_reroll.py
d5e255a5fdd06ea18e70f4077f855ed2f02664f91caf79b66bfbdf7df44ad07b tests/test_review_rehydration.py
@@ -203,11 +251,16 @@ cda09feb111efb8961e56c5f378f7925d63d0143b084c8248a9bc8be7f33ce6f tests/test_run
5f701a84f43a04c85d6f68627defd958e7dc9a1c0023632265dd2b1d19635b87 tests/test_runtime_redesign_r4_coordinator.py
4b03e1a9140ba32b3c24e1a2567ccbe90ecbe54f1b4e1ed760bc45da5fe9fb5b tests/test_runtime_redesign_r5_transactions.py
fdaf52fbcaea3f85f4e568f4fbb217d33069d6de8b5537bd97d53fcbc3e169a0 tests/test_runtime_redesign_r6_node_facade.py
d4c53d2c260b635d70fa211029d608f63776e0a56c20253475fa47c58363f8a7 tests/test_state_io.py
dc2f919e8540ad20272cc0bb45072e301a6d7b59e1aca22a5df9d6cf6f5cd720 tests/test_spectrum_portability.py
5af6a16a2337e3483bb0f9197fe04756decd309e4737520c5557b527d68d51bb tests/test_state.py
d4c53d2c260b635d70fa211029d608f63776e0a56c20253475fa47c58363f8a7 tests/test_state_io.py
1bbc7e5755f71d6fdfddcc6dc8337444febee87cb86773835ca6ac6c79d89357 tests/test_storage_api.py
82f049d7e51c8d6c43322ac4e0043d3fee6323c7989a2c446b86a936bd5ce790 tests/test_storage_frontend.py
a1f60f01e9d86c1f2a010752612e373e85112c7054ffc31185b9b459eb2b34a5 tests/test_temporal.py
8a512b3afdf394af043a4d6bd7c4fca0c9326e1207930bf5d96b4d0bd7205b4e tests/test_temporary_storage.py
7f518caa4d95fefd9794b7dafdb8b747963c98dad8473be73a00dff65f3c8ea4 tests/test_terminal_merge_diagnostics.py
f175e9ebf39d6fcaac578b640422676d7f7a54bfdf0f77ec44a3da6aafefd7cd tests/test_timeline_video.py
9388d33683b63dae2b51955f154d962237ddc93483437fa7c4e3c3165f17dde4 tests/test_v217_stable_facade.py
dc8b9cf0e77b4983ee4eecc60f4acde7d0e063f3b7569402feca8a56165107fd tests/test_v2_decoder.py
3cc94a6a7957ef285f58d373fbab001d926d2b43a754d60bb00bfce98675f32d tests/test_v2_prompts.py
424a4b34fb391ff3729b38a9b202e36f5bc8b9689a2a9bbf9589254c6261d9c6 tests/test_v2_registration.py
@@ -217,15 +270,13 @@ b42bcc0983147462d96c6b71cc4a38e846995dafa6eebefc3ae0d9bb36caf359 tests/test_v2_
0b5b20a8318862d167f135720e0489b8d4f93406d67c6646e052e35b03dcb876 tests/test_v2_session.py
b5c320279fd38d2038e39951a3d016aa9b80741523d2816e01ceda1bdecb71b8 tests/test_v2_terminal_merge.py
e22486ac62c53132c275614e4e161c78e6dce09dd0d1a7cf64f641a3b7052b99 tests/test_v2_widget_migration.py
c5b91ce02d46697be97920883f8c04b73871144210419f84b9c097c91bf41585 tests/test_v217_stable_facade.py
473abac48a57aa5f297c6e141a94de17ddcb9424a8744bae1c46e6a6a734a0ab tests/test_v3_hardening.py
9242f74663ccb04020ad2fb1199990c84433c9b4c5cf0560e93416c7b15e1913 tests/test_v3_physical_decode_groups.py
35b26c8cade7acb2eec35375d7dfa9a4a112f67a5cec6a842d61c69039de0d84 tests/test_v3_video_seam.py
1c84723b0c14f3169760fe1ae95d230416f6e947ada7d97ee890a2da33e9c988 tests/test_v31_run_storage.py
3af78ab85c8e19dee8d0a98bfd11a2c0e88c30d87256da1f9699c49a19a684e9 tests/test_v32_reference.py
43aef01fc587c9f9e87baf312eb4d588975aaaacab161a36bfcb638a4c11e31c tests/test_v34_driving_audio.py
aa1f36601489c360d2f1f989c10c5424bd85bd2115df2427afe696758eefd239 tests/test_v34_hybrid_reference.py
f457c3956fc804816783005a2349889796aeb5a6b930f09db0c3f43e051c6e2e tests/test_v35_compact_ui_frontend.py
3e9a79079823bd85b0e8848c8421a71ae7aa6ad2e5d47c05c7a5cc12ef9b2101 tests/test_v351_reference_audio.py
1c317c3443435e71bc3d874d1f1ef80aca6c50263ad5e9e393923ccadd3758f7 tests/test_v351_reference_audio_ui.py
3caacf5234d96b42d708c9a742d1a7f3b4318e208358058156f97cb21b5506ed tests/test_v35_compact_ui_frontend.py
75dbdd0d9694d82fab7da23f5cc03bc118c024abc865f6c282799aa8c57427fc tests/test_v35_conditioning_bridge.py
f9e191b2262030f988b6230e2a1f32484df03568fa43ac46d2c34adb61b6dffa tests/test_v35_context_sampler.py
7380c2ea1eeb4fb212ef51ffdf6d14a33132d3ae6d4c7ed5015f28bba6b7e75b tests/test_v35_hires_fix.py
@@ -235,8 +286,6 @@ f9e191b2262030f988b6230e2a1f32484df03568fa43ac46d2c34adb61b6dffa tests/test_v35
8ae5173eaec35adf9efcde2c464c95651c848248bdbdd6396ed2e3c3c1ac7965 tests/test_v35_refine_sampling.py
98a9006ebc4cab2b61b38904e0043fab24e088a25b28ffd38012c29991b8550c tests/test_v35_second_pass_contract.py
a11956a22f373d96f13b395971f39187d6bb12cdc153ce794b4729d258ce9725 tests/test_v35_second_pass_sampling.py
1c317c3443435e71bc3d874d1f1ef80aca6c50263ad5e9e393923ccadd3758f7 tests/test_v351_reference_audio_ui.py
3e9a79079823bd85b0e8848c8421a71ae7aa6ad2e5d47c05c7a5cc12ef9b2101 tests/test_v351_reference_audio.py
5f70cf8bd5248120ffcbe75ff3a54e23030fa6eb318ab0de6e0b08ee20e25a87 tests/test_v36_chunk_duration.py
bf69d1ac3fc30f963b6fed6cbd661f273593904b03275467427a954fbdc0a03f tests/test_v36_masked_video_prefix.py
19631b4abfdf2bd3c187ba4febbd23dd2bc6a0c5027f22c4adb8c1c4f1328926 tests/test_v36_public_sampler.py
@@ -255,33 +304,41 @@ bb5a1530c890d357dc48ea74039340d746e876e8f85e06d4d9a19369f8ff5961 tests/test_v38
8fb637f1d61389a86ef545adf35a518a5ae44322655a38fe598fd3b3c8e0a4a9 tests/test_v38_multi_reference_audio.py
623d528a02b6712cba28c374ba3ad979bc84e2ac5318640692589a04ba6ab15f tests/test_v38_open_integration_contract.py
12337a491fa67b38b32aaeb16e0b209393960dbc46db2a8bf968b8c3902855ef tests/test_v38_packed_row_planner.py
94eb249f407bf6858d936721d4a8752c5a95be6fa00dfc7f828b944e53ab1bc7 tests/test_v38_public_surface_cleanup.py
188ec8e125b16d0bec30b63324d23501a3b89578b434f8e7b24a9d8cc43f1640 tests/test_v38_public_surface_cleanup.py
7aebd3668e04233f34c2d94e75d4708cadd64d6882cf77ec59afa0a394dcb006 tests/test_v38_public_workflow.py
14ff6561a127287d0a20887790a978612ef9d291f1d3be7d5e85b37b2c74760e tests/test_v38_ref_encode_cache.py
8b376f45261d045b584f02b4d14cd70c2d6269b9c814d4e62b5c612f6e30e36f tests/test_v38_refine_target.py
03eb5bf6d845b22334fcc800d81b419f1c8d8593fa65a1e0460144456ff2c8ba tests/test_v38_reliability.py
aa72293fe5ba6f4499e1ec95aace3b2e94813fa35cf987a767e77bbf2ae525b1 tests/test_v38_resolution_preset.py
2ac395009e49f9554af8d1a251d93fc5d5e7070bf80f5a0b7b3f7e00cde15077 tests/test_v38_review_cache.py
975af1d6dd34e3403b80d3e89d8fc658baa67a34086d40cd05ac10fd92611c51 tests/test_v38_review_cache.py
72a0e3cb334049b66dd36f52c826931067c242fa63557933fd667138b38d6e8d tests/test_v38_review_control.py
f62daa235117a3041c66ebb17adedb13c0452c3a2fc0acdd51006d7aa0078a9b tests/test_v38_review_execution_cap.py
3a5e298bc9b0d46e9560e397576352117d6353dd03977ad54fe41e4881c58287 tests/test_v38_review_run_storage.py
2d8b74a56da4c15509ef9ef082f87321182d9d97ca6c417d5a6c055d67a4ee8e tests/test_v38_review_ui.py
259142da88241424afb61228b24306c88c1f99f0924e81595824ba09527e1847 tests/test_v38_review_ui.py
75697036f361a9b8ea74f69b15e408670f686c8c0e8a6c1d48fe059cc837d12a tests/test_v38_selective_second_pass_node.py
8026a241301dfc4de518649305458a538ef17a0b1567b113877a2af377738528 tests/test_v38_temporal_window_refine.py
0b2b1280a1217816984a99c0f948a52a8e2e6fdf924b5c4f2f578719b12b01ed tests/test_v38_video_audio_refine.py
68426689006a4bcdf3ae5b01589abf95c7fd3bf7f541904240412b24c0783fb0 tests/test_v38x2_decode_cache_integration.py
37107bbefa1a8c92abfc286b873efddd9a5feda2d0455e97149cc35a611cabfe tests/test_v39_official_workflow.py
20aff5243985a16550d1084ce20bb35cbb91150290a3514773defc1572a6c3f1 tests/test_v39_official_workflow.py
d33a6b53e3d074d5c8a076395b050fa2b6f00e25f222c32bd78e89d3e0d2635e tests/test_v39_r0_fixed_prompt_cache.py
319258798ca198312cf684f369990924657d1af168a148ad34303af31c06a350 tests/test_v39_review_refine_scope.py
68e5ee459f87724712d450d063bb4409cb26dd112b3b62602f1b43c88a3c08fb tests/test_v39_rr_r1_fixed_identity.py
4a97449504c720b0b4846293beb857b48986b3d175215509c144ee1b52c34f23 tests/test_v39_rr_r2_reference_routing.py
32e25a0f4815f954ba3e64e2445309babc83b78a964e2a6a5287b1c7accb275e tests/test_v39_rr_r3_effective_plan.py
6ec7852560f4e0588fc0b5d34743195554475167b46606d3166fdb5fb97f9c4e tests/test_v39_rr_r4_reference_runtime.py
ddeb44c42fb3543945d908a66b0557260ce719c002afd5bae0d37794219a8841 tests/test_v39_rr_r4_sequence_integration.py
307993ea60e2265a77ddf709350f51a134fceff6e2b26dcffd893af62162262e tests/test_v39_rr_r5_reference_storage_contract.py
62843c5c4d6150f66f4b232d9ec3c121fef984adfd8f316660be751c7c99ebe7 tests/test_v39_rr_r6_public.py
efe2cc9452f25dfc92f0cbe062d034c74d6121e9671a2ef0a9bb7852aa6a2652 tests/test_v39_rr_r6_public.py
49e525a86e23b26afd1008193aac510812cd2ba330d6b5b52023ab93d662ae27 tests/test_v39_stability_r1.py
b27004f82a3c551926e7e8f5bfa140af1eb2a1cf39ee5660f1d5b2495cf474ed tests/test_v39_stability_r2.py
d2e59eae0a724626b0a4e361921643dd31cb3612ae1f0d1b8b355c9d42c06174 tests/test_v39_stability_r3.py
072091309b35ff5c204c49ed478d9cdcc39d316df15a6209f687b33d13bbf790 tests/test_v39_stability_r4_r5.py
6e3f8a8a0480cc0a87a0ba7d5f79fdc1407233b155d90ff4e4b75aa57988ad49 tests/test_v39_turbo_sample.py
473abac48a57aa5f297c6e141a94de17ddcb9424a8744bae1c46e6a6a734a0ab tests/test_v3_hardening.py
9242f74663ccb04020ad2fb1199990c84433c9b4c5cf0560e93416c7b15e1913 tests/test_v3_physical_decode_groups.py
35b26c8cade7acb2eec35375d7dfa9a4a112f67a5cec6a842d61c69039de0d84 tests/test_v3_video_seam.py
0e02f7db41458fab91e6c3395bdb864d1c033923a3874ff4082d5f85b8230d98 tests/test_video_adapter.py
b07b941082a947d9e938ff399ec8b77fcd29c1a222e0e6b0dedf5d8e7482ce84 tests/test_video_reference_modes_experimental.py
0dbe8a5f71c92f3523942343b40787104c8b96781461ba3ba82934a6bdbe5834 THIRD_PARTY_NOTICES.md
f218e1a93c42467c6057c5c1d04d57cc82f6331b97ec9257e4e2dfb9c831876a timeline_video.py
0bc369ee428b8a5bc8e370a81fbab427262a1fdb7805a78c442c91698d65efba tools/__init__.py
3d5f29d1e446b94eb16bb02fea98a171c1b29248cbb4de66339030e3a7cea1c3 tools/a2_audio_only_canary_node/__init__.py
@@ -290,20 +347,20 @@ f218e1a93c42467c6057c5c1d04d57cc82f6331b97ec9257e4e2dfb9c831876a timeline_video
ea0040b510f748bc69f0c8aa4cc6aa1893e292af620ddd6e76e7847dccac1f8c tools/audio_continuity_research_runner.py
2876dc0f2760d6bffb0138ab4d6ef84da6f2db2d726cc442d7a6f732b7bc2061 tools/g4_sampling_diagnostic_node/__init__.py
9062a3b9e2d03551c8e230a34779ccd74b7e9e38aa18d19939900f0072919f82 tools/g4_sampling_diagnostic_node/node.py
0184162bf563e15b2681096732064e265c83696294e248c8a6981beb8255b568 tools/gpu_diagnostic_node/README.md
3c17d031c7f5c2a2449a0fa1e4b955b71663b043f072455350d3d25ee8e1c779 tools/gpu_diagnostic_node/__init__.py
3f9a6fb0d2f78d9b5d771b400d02c1c111a65e55ca6991da0b9dbf641a474aec tools/gpu_diagnostic_node/capture.py
f6dcddedf39be823e58dcd034e0fd583116d81f2f827860f714cb11356b9a1a4 tools/gpu_diagnostic_node/continuation_policy.py
92be5f45e1b4b3acf466d2cff24dfa119f64cdbbd5d0e22972c0a7ea9a4e4822 tools/gpu_diagnostic_node/nodes.py
0184162bf563e15b2681096732064e265c83696294e248c8a6981beb8255b568 tools/gpu_diagnostic_node/README.md
5021c682783be19007ecb63775413a1272a246c128c26f314fdaff8acc2bc748 tools/gpu_diagnostic_node/soft_reanchor.py
892804c81c18249348775d233c08508c298edecabc9ff151883053f350c44906 tools/install_gpu_diagnostic_node.ps1
68e3862e52bae7e98ee0db29f9c138db3e21007bbf4e27f1e8407c900a172df4 tools/issue13_drift_analyze.py
e922c8737c8e4a1d62299d27391c0496ecec83c0e9f26600b2039f83b8272b18 tools/issue13_drift_gate_runner.py
653d7204e48edef6caec14b57a539b54fe6fd32cb5659c0d7c43726b711a4487 tools/issue13_r2_analyze.py
55caa8df2cd57289bf39c0b9a9a43440e7c3aa2e5bb89176cb5238c15884bd8d tools/issue13_r2_gate_runner.py
c315aacb3fbe6dc8a38c50cfb336760ee2fd3c293d80617360c3f3b639ae2053 tools/issue13_r25_gate_runner.py
f03916a76a1c1489fd7a23d234a04d0f26fa8afb49a409e0fea779b9fac71a11 tools/issue13_r28_analyze.py
ab4b007df62fc09a6d346dcb8c5171c8ea8b6e56e487720ef9fb91ddf19014d3 tools/issue13_r28_gate_runner.py
653d7204e48edef6caec14b57a539b54fe6fd32cb5659c0d7c43726b711a4487 tools/issue13_r2_analyze.py
55caa8df2cd57289bf39c0b9a9a43440e7c3aa2e5bb89176cb5238c15884bd8d tools/issue13_r2_gate_runner.py
2743597d289d9eb12334d1f5cdd630497e237814ddf48b17f6f68da299c01150 tools/p20_file_backed_tensor_poc.py
3bf6dfddb9a0e5276613e505465d0f01442035940c0806a3ca64daf7753a8592 tools/p24_comfy_integration_probe.py
6277e2aad67ddc9b446c9883a33e72b5194a053e5b814a9a935ce7a7770d8793 tools/p25_gpu_acceptance_probe.py
@@ -349,14 +406,14 @@ dc2b901789e4b024dbcbc4024c63eb8cbc564d77003ce8c35c0584851ba6dfbd v2/prompts.py
be0d8585fa57e0ccd6d02577bbc7299629e5bfe20e186747f571fccc98b67d7e v2/seam_types.py
12cfcc0c71ca087fe39a5ce2922959a62b2bb311bfe664f05c4480affce03708 v2/seeds.py
0de733c55bb61e69d48918eb5bc91318fc7cf3af383fb8937522ced9df018c47 v2/sequence.py
13213e7709fd02bb810af3c18e4cd7d010f8cdfb121a4b99866a9f2333331089 v2/session_io.py
3ea23dbce6fcd217b6e81235855479b9b16fed7a21693cf156f9014555c4fe13 v2/session.py
13213e7709fd02bb810af3c18e4cd7d010f8cdfb121a4b99866a9f2333331089 v2/session_io.py
78c830cdf49034c6e3b2cd3da5914d5cfc5638eb63ec6dbd1bab13c50d6d4caf v3/__init__.py
ea2922567df90436dd6fa21696833a49dd0cbfeee142a5fedc3a7c303824d473 v3/adaptive_continuity.py
70bec17d0dce1873b9bb4a51b09daf76833722ce89f005a22b029533a367e62b v3/assembly_v35.py
7815230e0a9540bc2edeaf818e940b7a1fdd817fd217ba8a4f9cb21aaf6a308e v3/assembly.py
70bec17d0dce1873b9bb4a51b09daf76833722ce89f005a22b029533a367e62b v3/assembly_v35.py
3f62fdb35e804f42d7cbf0828b4e87e69a2b99246c3f7467fe4fa698f1d803c9 v3/conditioning_bridge_nodes.py
fa7a355e52fbcfe1f53b16fa1c37af5d839215b49412adb41c90129ae4e9d9f7 v3/driving_nodes.py
6403c2bde3fc0b8cddef78567f663fddb48130628f4c9258fcd679c28fd0cf13 v3/driving_nodes.py
147a1d3fc5e09b323a4c3d13d30e432410389bae52b58330462375b9f307b791 v3/easy_audio_nodes.py
64b3e54b42e5156642f9957ccc374fd190d9fc7145637984e702534fb1bf80b4 v3/easy_image_nodes.py
ac859cb1a6e23b8ceae300d818c24fe1346b649a0d663017a9b3ac64aa96fca3 v3/easy_nodes.py
@@ -393,14 +450,14 @@ ae909b43ad17934c85881d3b0bea81e7b45c215eeb4d1f9f6a9ea72faf633ddf v3/reliability
3f03d8659305d2def41939205d21cdc6f35e32943274e87afe539d15b9635f9b v3/review_control.py
be6610fb9427546d9485a17015cfbdc8786715e6a0967f3e4c1dee35d8768385 v3/runtime_coordinator.py
d0a004aa3030c8adaa5ea788f9acc758fcb4a556f203f9a3f63fa4e4b5772063 v3/sampling_engine.py
452022107cea233aa84d5366ec4943ac5aadbd5fcf786113f1d542a641eac8f7 v3/second_pass_nodes.py
97fcb293e0714c3a74b89f08e11c21dce45d25a3010948be1af9dc059d288187 v3/second_pass.py
452022107cea233aa84d5366ec4943ac5aadbd5fcf786113f1d542a641eac8f7 v3/second_pass_nodes.py
6e71e944313339f14e14e11cec682593159ca5542aa5d38095371da781d60224 v3/targeted_refine_sampling.py
5a044bdd70c546e4446e0c26af92aae88049ffa2e0e2d01b52033a83a3152d0e v3/targeted_second_pass.py
26b13614ba0a5fd8b192a106c9503b2f4b6fe7f9268a0fcb0f544979a49cab34 v3/video_adapter_nodes.py
4178dcfa442abdb06ce276e05bd7bcbaa6da28469ba5e57d3da3c663de4e4a9d v3/video_seam_v35.py
2f07ace1462644f0bee553827b3571535de4d0c2d9ef847bd9e0ca73a4eed5cb v3/video_seam.py
c9e03d72565bac7639576474c258ba6fcb0fc73047bd9adb1292bb6ae6db8495 version.py
4178dcfa442abdb06ce276e05bd7bcbaa6da28469ba5e57d3da3c663de4e4a9d v3/video_seam_v35.py
e124ded4bc8ddbf7b432c80d9900e40ee34214327bc6d83bd9fcf95cb55e7169 version.py
b935ad3bb729948bb3890ce967b0ecbef737a5e51ddfc47a0060fc20aa78dcda video_reference_modes.py
685df8e0fa23e3663ab31da1e4470cd19bdaa0dc16362cb606e874d9d6153c2c web/decode_cache_helper.js
fefafd4550511d80bd85ade46a0476f0cf3de13ddb94a908dff6b175f2a0e199 web/easy_bypass_toggle.js
@@ -408,35 +465,13 @@ fefafd4550511d80bd85ade46a0476f0cf3de13ddb94a908dff6b175f2a0e199 web/easy_bypas
9db0aa94f634578a962d5db49e5c70356040a59df968e3a2e531b327a623081f web/easy_load_image.js
35090816ba7470a7255eb85d003b7e79f8a0065982dacf98a18f41c8ebff655d web/easy_load_video.js
0c68d13a1f8beb7431a32b471f8b0612c42fcc4bf72dd97611a025d158008662 web/easy_v38.js
90bef8b7628106fffa95ac1e27e345fa3ccc7203bd2f608497f46b751f9bb850 web/project_id.js
933b5c675294a20ccec263916a1f4453d5b06a530d660ad8e56389d02d7f3023 web/project_id.js
ae59273f163a116ece418957c557c4fd8c3285e3c60814679b9361d2e6e6fa02 web/reference_audio_ui.js
46c9f75c9462bf26300332448f4431cca801f38964ca0a49bc2c50a88a7cc40f web/reference_image_ui.js
beea70f6e25e1366aba5e2689c31733478472f207dc50cad1562cd877152e4f9 web/reference_images_v39.js
e99c74b78da1aa18be61cb053fa05e5abd106caa5eb5c814cf585786aae1e7f8 web/reference_images_v39.js
42afab2241d39c1b07ff33d8b1370d4b0f729d81e53b4c99f4bc54c6a96b8800 web/storage_settings.js
bf41bf08dd0504eb753fa5781e6c2b8c1285758ad4bb7309975bb9231d5f8e41 web/video_reference_mode_ui.js
2bcf455b4ece8b1861e327f6f9f8a79170e7c77470155aacada52e18438543dd write-h3-v39-reference-prompts.zip
3dfd0e7042070f62b9e78239c0e75d82a7df32098db10832465f3049b7f0b38f write-h3-v39-reference-prompts/SKILL.md
a9f7b67bd65a9beda622aa5fdc5f8a2ecae1b58b18690d91fb0107a16a3e384e write-h3-v39-reference-prompts/references/simple-3x5.txt
81d7b4f4da797339b46911bc96d0a3736d70df29e092715c6684bb1786908593 write-h3-v39-reference-prompts/references/structured.md
3dfd0e7042070f62b9e78239c0e75d82a7df32098db10832465f3049b7f0b38f write-h3-v39-reference-prompts/SKILL.md
7fd122ff08232e114a6ef32171217ea6324e980eb7257d560c466b8177a3d3ec audio_compat.py
82efab13baacb9a229671eda5f4a2d24bb839b160b2bfb40591df51a5ddf80a4 tests/fixtures/v39_sampling_v5/project.json
8b8ca1b0c802d024c6915d903747d8722bea25628369997ab008ff896462b96a tests/fixtures/v39_sampling_v5/revisions/d8352287737286e5/chunks/rev-d8352287737286e5-txn-008f52b93a88461481ead25f91627af0-chunk-0002.safetensors
26e771cffc57f171f4cc4c6859876350bdf717d1e8213fc005343006a34f002f tests/fixtures/v39_sampling_v5/revisions/d8352287737286e5/chunks/rev-d8352287737286e5-txn-9bb0a584b7344342b114bd31b175225c-chunk-0001.safetensors
7bf0cf46f049e072a68e71cf8bbab10aef695f0931b16842a890c084021f5a11 tests/fixtures/v39_sampling_v5/revisions/d8352287737286e5/manifest.json
49e525a86e23b26afd1008193aac510812cd2ba330d6b5b52023ab93d662ae27 tests/test_v39_stability_r1.py
b27004f82a3c551926e7e8f5bfa140af1eb2a1cf39ee5660f1d5b2495cf474ed tests/test_v39_stability_r2.py
d2e59eae0a724626b0a4e361921643dd31cb3612ae1f0d1b8b355c9d42c06174 tests/test_v39_stability_r3.py
072091309b35ff5c204c49ed478d9cdcc39d316df15a6209f687b33d13bbf790 tests/test_v39_stability_r4_r5.py
319258798ca198312cf684f369990924657d1af168a148ad34303af31c06a350 tests/test_v39_review_refine_scope.py
8d0ccf5711e1be0bf63b4e02fecb0b37cd88979894d27b89337cda4a6d1c4ec5 tests/frontend_issue30_ui.cjs
d6b5b969a0c02f1b282da990db3e503e0768747cfac8a4edd32cf8677550e024 tests/test_issue30_frontend.py
4546ffb0cbaf270c52d1900908e46bfa2d3c0e2d0035ae5694680890e89aedd8 docs/TEMPORARY_STORAGE_JA.md
660a9df28c7d62ad383cda5514c48c7efcb3576f435af023040116d533a56563 storage_api.py
517662b104ab1507bb927a39f25b8503f092c2ba4c39bc277379ba9d9e31b334 storage_fs.py
daeda14d9f5338835661e6738e074a3897d936cd847fe3e78519efabbcdfd0ef storage_memory.py
53da3ac42f4a66e8adc4560049c7c8e7ada82599be8c61f0573c234342ca36ed storage_settings.py
ea47f51abe07e9b2af840272939c0f363369636af30f4c6dffb0bb696b752c8a tests/frontend_storage_settings.mjs
1bbc7e5755f71d6fdfddcc6dc8337444febee87cb86773835ca6ac6c79d89357 tests/test_storage_api.py
82f049d7e51c8d6c43322ac4e0043d3fee6323c7989a2c446b86a936bd5ce790 tests/test_storage_frontend.py
8a512b3afdf394af043a4d6bd7c4fca0c9326e1207930bf5d96b4d0bd7205b4e tests/test_temporary_storage.py
42afab2241d39c1b07ff33d8b1370d4b0f729d81e53b4c99f4bc54c6a96b8800 web/storage_settings.js
3b45e4b58290504366030988fc4e0294ab6428205d520405c9fd1cb9243efa4f .github/workflows/publish_registry.yml
+24 -3
View File
@@ -1,10 +1,31 @@
ComfyUI-H3-Continuum 3.9.4 / 一時ストレージ設定
パッケージmetadata: 3.9.4(2026-10-13)。公開登録は12ノード。
ComfyUI-H3-Continuum 3.9.9 — public KJ / Review continuation
CPU 2075 passed / 2 skipped / 2 known xfailed; upstream KJ GPU and installed local-KJ browser Continue PASS.
既存Take・Workflowは維持。更新後は再起動・ブラウザー再読込。旧unsafe/v5は自動変換しません。
対応条件と検証範囲: docs/RELEASE_NOTES_V399.md / docs/KJ_PREVIEW_OVERRIDE_V398.md
**最新版パッケージは3.9.4です。** 固定版のGitHub配布は下表を参照してください。Registryの公開状態は別に確認します。
--- Historical package record (unchanged) ---
ComfyUI-H3-Continuum 3.9.8 / KJ Model Preview Override continuation
Full real-Core CPU gate: 1,995 passed / 3 skipped / 2 existing Inspector expected failures / zero unexpected failures and CUDA initialization attempts; the 48 dedicated KJ tests passed without skips. Normal ComfyUI GPU acceptance passed four Core-API Queues: PyTorch and Comfy Kitchen INT8 + the audited local Spectrum implementation, each with separate-Queue Chunk 1 to 2 continuation and held preview images. Both second Queues reused one chunk and generated one; original Chunk 1 bytes and contracts were unchanged. All four video/audio outputs fully decoded. See docs/KJ_PREVIEW_OVERRIDE_V398.md for exact support and limits.
Support: explicitly audited provider only; Preview Override VAE/Audio VAE disconnected, Tiny VAE none, Core NoPreviews/Auto/Latent2RGB. Restart and start a new Run. Old unsafe/v5 Takes and workflow bytes remain protected.
--- Earlier package validation records (historical) ---
ComfyUI-H3-Continuum 3.9.6 / Regeneration intent and automatic variations
Full CPU gate: 1,908 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and CUDA initialization attempts. Normal ComfyUI browser save/reload retained Chunk 2 / nonce 5 and Chunk 1 / nonce 0 after history loading. Five new-Run GPU queues passed continuation, Chunk 2 regeneration and two identical Full Run / nonce 0 submissions; earlier Chunk 1 bytes were preserved. GPU scope: PyTorch, Spectrum OFF, 512x512, 8-step Turbo; all five media outputs fully decoded. See docs/REGENERATE_INTENT_V396.md.
Update: restart ComfyUI and refresh the browser; select and save old unmarked Chunk 1 / nonce 0 again. Do not delete Runs/Takes. GitHub Release/Registry upload, service scan approval and Manager installation are separate publication checks.
--- Previous 3.9.5 package record (historical) ---
Previous 3.9.5 validation: 1,893 CPU passed / 2 skipped / 2 existing Inspector xfailed / failures 0; CUDA initialization 0. Public Spectrum 0.2.29 PyTorch/Kitchen Chunk 1→2 GPU continuation and full 10-second AV decode PASS. Details: docs/ISSUE31_SAMPLING_SIGNATURES.md
ComfyUI-H3-Continuum 3.9.5 / Issue #31 保存Take継続
パッケージmetadata: 3.9.5(2026-10-15)。公開登録は12ノード。
**最新版パッケージは3.9.5です。** 固定版のGitHub配布は下表を参照してください。Registryの公開状態は別に確認します。
| バージョン | Release | カスタムノード本体ZIP | SHA-256チェックサム |
| --- | --- | --- | --- |
| 3.9.5 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.5) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/ComfyUI-H3-Continuum-v3.9.5.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/SHA256SUMS.txt) |
| 3.9.4 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.4) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/ComfyUI-H3-Continuum-v3.9.4.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/SHA256SUMS.txt) |
| 3.9.3 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.3) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/ComfyUI-H3-Continuum-v3.9.3.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/SHA256SUMS.txt) |
| 3.9.2 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.2) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/ComfyUI-H3-Continuum-v3.9.2.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/SHA256SUMS.txt) |
+30 -3
View File
@@ -1,13 +1,40 @@
ComfyUI-H3-Continuum 3.9.4 / Registry packaging repair - 2026-10-14
ComfyUI-H3-Continuum 3.9.9 — public KJ / Review continuation
CPU 2075 passed / 2 skipped / 2 known xfailed; upstream KJ GPU and installed local-KJ browser Continue PASS.
既存Take・Workflowは維持。更新後は再起動・ブラウザー再読込。旧unsafe/v5は自動変換しません。
対応条件と検証範囲: docs/RELEASE_NOTES_V399.md / docs/KJ_PREVIEW_OVERRIDE_V398.md
Package metadata: 3.9.4. Public registration: twelve nodes. Existing V3.9
--- Historical package record (unchanged) ---
ComfyUI-H3-Continuum 3.9.8 / KJ Model Preview Override continuation
Full real-Core CPU gate: 1,995 passed / 3 skipped / 2 existing Inspector expected failures / zero unexpected failures and CUDA initialization attempts; the 48 dedicated KJ tests passed without skips. Normal ComfyUI GPU acceptance passed four Core-API Queues: PyTorch and Comfy Kitchen INT8 + the audited local Spectrum implementation, each with separate-Queue Chunk 1 to 2 continuation and held preview images. Both second Queues reused one chunk and generated one; original Chunk 1 bytes and contracts were unchanged. All four video/audio outputs fully decoded. See docs/KJ_PREVIEW_OVERRIDE_V398.md for exact support and limits.
Support: explicitly audited provider only; Preview Override VAE/Audio VAE disconnected, Tiny VAE none, Core NoPreviews/Auto/Latent2RGB. Restart and start a new Run. Old unsafe/v5 Takes and workflow bytes remain protected.
--- Earlier package validation records (historical) ---
ComfyUI-H3-Continuum 3.9.7 / External-input regeneration cache
CPU: 1,948 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and zero CUDA initialization attempts. Targeted Core-API GPU queues verified external Chunks=2 / Nonce=0, repeated identical automatic variations 1 then 2, and 1 reused / 1 generated chunk per variation. Browser save/reload and Run-button submissions were checked separately. GPU scope: PyTorch attention, Spectrum OFF, 512x512, 8-step Turbo, two 5-second chunks. This is a targeted cache repair, not general image-quality or all-accelerator acceptance.
Existing workflow and Take files were hash-protected. No storage/schema migration.
--- Previous 3.9.6 package record (historical) ---
ComfyUI-H3-Continuum 3.9.6 / Regeneration intent and automatic variations
Full CPU gate: 1,908 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and CUDA initialization attempts. Normal ComfyUI browser save/reload retained Chunk 2 / nonce 5 and Chunk 1 / nonce 0 after history loading. Five new-Run GPU queues passed continuation, Chunk 2 regeneration and two identical Full Run / nonce 0 submissions; earlier Chunk 1 bytes were preserved. GPU scope: PyTorch, Spectrum OFF, 512x512, 8-step Turbo; all five media outputs fully decoded. See docs/REGENERATE_INTENT_V396.md.
Update: restart ComfyUI and refresh the browser; select and save old unmarked Chunk 1 / nonce 0 again. Do not delete Runs/Takes. GitHub Release/Registry upload, service scan approval and Manager installation are separate publication checks.
--- Previous 3.9.5 package record (historical) ---
Previous 3.9.5 validation: 1,893 CPU passed / 2 skipped / 2 existing Inspector xfailed / failures 0; CUDA initialization 0. Public Spectrum 0.2.29 PyTorch/Kitchen Chunk 1→2 GPU continuation and full 10-second AV decode PASS. Details: docs/ISSUE31_SAMPLING_SIGNATURES.md
ComfyUI-H3-Continuum 3.9.5 / Issue #31 Attention and Spectrum identity - 2026-10-15
Package metadata: 3.9.5. Public registration: twelve nodes. Existing V3.9
and separate V3.8X2 workflows remain supported without schema migration.
Release artifacts and integrity (2026-10-13)
**Latest package: 3.9.4.** Use the fixed-version GitHub downloads below. Registry publication is verified separately.
**Latest package: 3.9.5.** Use the fixed-version GitHub downloads below. Registry publication is verified separately.
| Version | Release | Custom-node ZIP | SHA-256 checksum |
| --- | --- | --- | --- |
| 3.9.5 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.5) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/ComfyUI-H3-Continuum-v3.9.5.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/SHA256SUMS.txt) |
| 3.9.4 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.4) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/ComfyUI-H3-Continuum-v3.9.4.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/SHA256SUMS.txt) |
| 3.9.3 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.3) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/ComfyUI-H3-Continuum-v3.9.3.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/SHA256SUMS.txt) |
| 3.9.2 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.2) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/ComfyUI-H3-Continuum-v3.9.2.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/SHA256SUMS.txt) |
+83 -2
View File
@@ -1,13 +1,94 @@
# Public source status
Updated: 2026-10-14 (Japan time).
## 3.9.10 — Spectrum reuse and Reference assignment display
Normalize Spectrum CRLF source auditing and add fixed Python 3.12 profiles while
preserving the existing 3.13 profiles, valid Sampling v6 identities, old Takes
and official workflows. Correct the Reference assignment display for externally
connected Chunks without changing generation count or saved selectors.
Full local CPU/JS: 2,096 passed / 2 skipped / 2 known expected Inspector failures.
Python 3.12 component: 112 passed / 1 skip; 24 independent storage cases and
actual-JS checks passed. Reference browser checks passed. No new GPU acceptance
is claimed. Restart ComfyUI and refresh the browser; use a new Run for newly
supported configurations. See docs/RELEASE_NOTES_V3910.md.
## 3.9.9 — KJ Preview Override / Review continuation
Fix new-Run v6 read-only with the separately audited upstream KJ profile, and
show saved reasons when reuse is unavailable. Keep the original local profile,
Sampling v6 guards, old Takes and official workflow bytes.
CPU: **2,075 passed / 2 skipped / 2 known expected failures**, plus overlapping
90-case JS coverage. Upstream KJ GPU and normal-ComfyUI browser Continue both
passed: **Chunk 1 preserved, Chunk 2 generated, one reused / one new**; all four
AV outputs fully decoded. Browser checks used the existing local KJ profile.
These are targeted repair checks, not general quality or all-provider acceptance.
Restart ComfyUI and hard-refresh after updating. On Preview Override, disconnect
optional VAE/Audio VAE and set Tiny VAE `none`; retain generation VAEs. Correct
unsupported settings before starting a new Run. Old unsafe/v5 Takes remain
readable and are not automatically made reusable. No workflow rewiring or Take
deletion is required. [Full release notes](docs/RELEASE_NOTES_V399.md).
## V3.9 Turbo Sample on main
Add the user-supplied sample separately: Auto, 2 x 8s, multiline image-anchored
dance prompt, fixed background/camera, Thai rap and one enabled Turbo LoRA.
Preserve the supplied 6-step/Draft settings and document their unverified pairing
with the 4-step/768p LoRA. Fresh workflow/project IDs avoid user Run identity reuse;
remove misleading auto-download metadata for the saved custom model. Standard
workflows, existing Takes and runtime code stay unchanged. No new GPU validation,
runtime deployment, release or Registry publication is included.
Updated: 2026-10-16 (Japan time).
## Japanese README organization (documentation only)
The Japanese README follows the English quick-start structure. Japanese detail
and images remain in docs/USER_GUIDE_JA.md and docs/README_HISTORY_JA.md.
Downloads, model folders, provider links, history safety and limited acceptance
scope are retained. English documentation, runtime and published assets are unchanged.
## README organization (documentation only)
The English README is a compact entry point. Detailed operating instructions and
screenshots remain in docs/USER_GUIDE.md; release and older implementation notes
remain in docs/README_HISTORY.md. Version-specific acceptance statements retain
their historical context. Japanese README, product code, official workflows and
published v3.9.8 assets are unchanged; this is not a new runtime or release version.
## Published v3.9.8 GPU recheck (2026-10-16)
The exact published tag `58fbd855` was deployed and restarted in normal ComfyUI; all 464 release files and the live startup/adapter identity matched. Two new Runs / four Core-API Queues passed PyTorch/KJ and Kitchen INT8 + audited local Spectrum/KJ Chunk 1 to 2 continuation, unchanged prefix/contracts and complete AV decoding. Existing Takes/workflows stayed hash-identical. README and docs/RELEASE_NOTES_V398.md record conditions and unverified scope. The later documentation-only main commit does not replace the release tag, GitHub assets or Registry package.
## 3.9.8 — KJ Preview Override Phase A (2026-10-16)
A bounded adapter identifies the audited KJ Preview Override source and loaded helpers, preserves fixed settings and wrapper execution order, and excludes its reviewed transient image list. VAE/Audio VAE must be disconnected, KJ Tiny VAE must be `none`, and Core approximate-VAE preview is outside this phase. Unknown implementations remain non-reusable. Sampling v6, old Takes and official workflows are unchanged. Full real-Core CPU gate: 1,995 passed / 3 skipped / 2 existing Inspector expected failures / zero unexpected failures and CUDA initialization attempts; the 48 dedicated KJ tests passed without skips. Normal ComfyUI GPU acceptance passed four Core-API Queues: PyTorch and Comfy Kitchen INT8 + the audited local Spectrum implementation, each with separate-Queue Chunk 1 to 2 continuation and held preview images. Both second Queues reused one chunk and generated one; original Chunk 1 bytes and contracts were unchanged. All four video/audio outputs fully decoded. See docs/KJ_PREVIEW_OVERRIDE_V398.md for exact support and limits. Release and Registry state are verified separately.
## 3.9.7 — external-input regeneration cache
Fix automatic Full Run variations when Chunks or Variation Nonce come from external nodes. Unresolved linked inputs no longer let Core skip an explicitly requested regeneration. Earlier accepted chunks, existing Takes, public inputs and official workflows are preserved. Update the custom nodes, restart ComfyUI and refresh the browser; no workflow changes or history deletion are required. See docs/RELEASE_NOTES_V397.md for targeted validation and limits.
## 3.9.6 — regeneration selection and automatic variations
Package 3.9.6 preserves saved regeneration selections and routes explicit Full Run automatic variations through Run Storage on every Queue. The full CPU gate and targeted normal-runtime browser/GPU acceptance passed; scope and update steps are in docs/REGENERATE_INTENT_V396.md. Sampling v6 safety, public inputs, official workflows and stored Takes are preserved. Release assets and Registry service status are verified separately.
## 3.9.5 — Issue #31 Attention and Spectrum identity repair
Finite loaded-implementation signatures identify the Core PyTorch/Kitchen attention overrides and explicit Spectrum profiles, including upstream 0.2.29. Existing normal v6 contracts without the targeted override remain unchanged; original unsafe/v5 history and official workflows are preserved. Validation and compatibility limits are recorded in docs/ISSUE31_SAMPLING_SIGNATURES.md. GitHub assets and Registry service status are verified separately.
## GitHub and Registry publication status
**Latest package: 3.9.4.** Use the fixed-version GitHub downloads below. Registry publication is verified separately.
**Latest package: 3.9.8.** Use the fixed-version GitHub downloads below. Registry publication is verified separately.
| Version | Release | Custom-node ZIP | SHA-256 checksum |
| --- | --- | --- | --- |
| 3.9.8 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.8) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/ComfyUI-H3-Continuum-v3.9.8.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/SHA256SUMS.txt) |
| 3.9.7 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.7) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/ComfyUI-H3-Continuum-v3.9.7.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/SHA256SUMS.txt) |
| 3.9.6 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.6) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/ComfyUI-H3-Continuum-v3.9.6.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/SHA256SUMS.txt) |
| 3.9.5 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.5) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/ComfyUI-H3-Continuum-v3.9.5.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/SHA256SUMS.txt) |
| 3.9.4 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.4) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/ComfyUI-H3-Continuum-v3.9.4.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/SHA256SUMS.txt) |
| 3.9.3 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.3) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/ComfyUI-H3-Continuum-v3.9.3.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/SHA256SUMS.txt) |
| 3.9.2 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.2) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/ComfyUI-H3-Continuum-v3.9.2.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/SHA256SUMS.txt) |
+284 -1609
View File
File diff suppressed because it is too large Load Diff
+250 -737
View File
File diff suppressed because it is too large Load Diff
+61 -40
View File
@@ -1,7 +1,15 @@
ce319f06573b0b53709bae56d9b9ee0b3f4180eba961d7cbbac9e01e403a589e .comfyignore
fa9c99303d1a960db12ae2191cffcfbffe82e25e144eb6d16775c840fec4d573 CHANGELOG.md
fe29dafce4a6c5229e7e3aef1f946afba2346015702ed8cdda9927a80b6e65f3 INSTALL_JA.txt
c44abbd789091bbead01250dbd5453e60539d2665c69f162335fdbd2936019c2 LICENSE
656e417909695c5dcabf134b6f3240c44bd02bb2e33881b5247efd345cb932c2 PACKAGE_INFO_JA.txt
57fa9ad77ab5b1d2da6ec43f7ba19118e253cb4a959b758b246755471df15591 PACKAGE_VALIDATION.txt
da45eeb9f4950777bb63974cbf1d77856b287b05e3466be63571f2ad4561f12e README.md
eed26374c7de5e9e19396836e0270fb46f37f3302dfd6f5533c8d703741cbf0f README_JA.md
0dbe8a5f71c92f3523942343b40787104c8b96781461ba3ba82934a6bdbe5834 THIRD_PARTY_NOTICES.md
02ea0951ce2e7ac2d5f3bbf02c596a9f9ea8adcdf2a113dca4c8db3cc19c36b3 __init__.py
48c25e689d401e4764fea0084100cdd43f8ff524410d10a9be49d3befb8b7202 .comfyignore
7fd122ff08232e114a6ef32171217ea6324e980eb7257d560c466b8177a3d3ec audio_compat.py
3e7a1ec3380be2a222c8a3e9b8a6bb184d3a760478957f48186438f15235bdba branch_provenance.py
1358bcdd91ba8c4ee29c92c203bcdea9c34fd0ad7437bc0f00b657ce0ae32c72 CHANGELOG.md
c1ed3a0342730450958fb62fd21854a876991f3a4aefb24e5383b0c73cd2312d compatibility.py
4d62508f900af9f93ec320afb900c27cc9a300d1e0c0a94714258498c5ede5fc conditioning.py
294251c40b1be2e296c5e5f627b68091d087bdb39179e3a0c3cebd2ee8a344f5 constants.py
@@ -11,55 +19,74 @@ cf255b8a635471229aebadc926acfdcacb65963adbcb9e0f9505d04eada2c6b5 continuation.p
c2b44e3caed54c894556af547cd800328ed287a8cc90e0b5031f34bd6bd2224b decode_cache/nodes.py
986cc7a3dc153ff1b41c427a86756e505ea6e17d8b065df1ab15b8ff904f090e decode_cache/service.py
630ba3d2183964444c4341a8df507f7af61dac6a8e5aaca8cf387a5376d49f27 decode_cache/store.py
44377cbac9131daa3d1c88ccc154c325257767f4c52f3181f61525c7eae00531 docs/ISSUE31_SAMPLING_SIGNATURES.md
b7ecd069d803cae7f86ad0ab89df144b5a5b157c168c73ee2962b47234d4ec0c docs/KJ_PREVIEW_OVERRIDE_V398.md
77c225d1e49e06d00b04acf811177c9a4317e792d4bfe92df539de99420c3563 docs/LOADER_PERSISTENCE_REPAIR.md
b30fe4640359bebfcdeeb94dbeec94cb346c8e3646509e1dcdb87d7021ba04ff docs/N2B_BRANCH_PROVENANCE_CONTRACT_V1.md
bce0337cce903938b6b8df7f52c5c67e2d0ee14599c693687e4180debc606db2 docs/README_HISTORY.md
9ee92cdd3783e61e3bc71b99d5c0df7dd0a9f129c8a6b37701d7001a4f745e77 docs/README_HISTORY_JA.md
cebf8462681796df807987d8afef2c78abb0953b54b2f5dcd12d583bde0ec806 docs/REGENERATE_INTENT_V396.md
bbe74f802339b758d8874900e2617dd1a46844a58e0f872d95f34b9b40ba20ed docs/RELEASE_NOTES_V3910.md
49d02b7650d4a542307c9d4189dc217ba1db54c4eac959cc2e2a1ce57b501ae3 docs/RELEASE_NOTES_V399.md
91ee4adf828415e84e2f413e3a4e20dc16356f139e8ef1f6756f584cad04f3c8 docs/REVIEW_RESTORATION_REPAIR.md
4546ffb0cbaf270c52d1900908e46bfa2d3c0e2d0035ae5694680890e89aedd8 docs/TEMPORARY_STORAGE_JA.md
4c33d223a77c78e5c74946aa43f537031fd5aa415388d46fed975f9b07d53c3e docs/TIMELINE_VIDEO_EXPERIMENTAL.md
00f88cb57be5e79d9c881671bd13fa154d91843f4c1ae2f2451ee2a73841271d docs/TURBO_SAMPLE.md
92bee3f3c0c340356abdd06b4993fbf2e98cbe57e931f85cbfd11da0f7f87809 docs/USER_GUIDE.md
c3f0753bdae1c54044a6e9ad5770c45e85a306ac9c15cd363790f8625c865f94 docs/USER_GUIDE_JA.md
717577efdec5e61600504b77bc2a099ef022c4fe2333e936c3310c7182d8f2a0 docs/V38_OPEN_INTEGRATION_CONTRACT.md
3496636f645b39336fd95a4f21a562810c56f8e1d92fa6eeb1e4b0786ad11b37 docs/V38_RELEASE_AND_MIGRATION.md
4d17d74107e78404b190fec9494b443249e5aff4f58670c50d54dfdfb064334b docs/V39_REFERENCE_INLINE.md
a4705ae4b5730d2b43e29fe125a64196a031f14bcfa8702663a17aa89f683d89 docs/images/v39/v38x2-feature-summary-historical.png
ae3338f720fc22c05ab4a661c68f9c5dcb20d751a835f0e60d4738333be48e63 docs/images/v39/v39-feature-summary.png
84304d318eea6db3921d366647c6fff9ad488e8c4cfef6ca00988bfd2a6504bc docs/images/v39/v39-reference-images-node.png
b765b37502b012b462606c4d2dcef43f13950e5fd682824fb18cdb19d542c121 docs/images/v39/v39-sampler-node.png
77c225d1e49e06d00b04acf811177c9a4317e792d4bfe92df539de99420c3563 docs/LOADER_PERSISTENCE_REPAIR.md
b30fe4640359bebfcdeeb94dbeec94cb346c8e3646509e1dcdb87d7021ba04ff docs/N2B_BRANCH_PROVENANCE_CONTRACT_V1.md
91ee4adf828415e84e2f413e3a4e20dc16356f139e8ef1f6756f584cad04f3c8 docs/REVIEW_RESTORATION_REPAIR.md
4c33d223a77c78e5c74946aa43f537031fd5aa415388d46fed975f9b07d53c3e docs/TIMELINE_VIDEO_EXPERIMENTAL.md
717577efdec5e61600504b77bc2a099ef022c4fe2333e936c3310c7182d8f2a0 docs/V38_OPEN_INTEGRATION_CONTRACT.md
3496636f645b39336fd95a4f21a562810c56f8e1d92fa6eeb1e4b0786ad11b37 docs/V38_RELEASE_AND_MIGRATION.md
4d17d74107e78404b190fec9494b443249e5aff4f58670c50d54dfdfb064334b docs/V39_REFERENCE_INLINE.md
a8c56379abe23b98c8ceed6ea5e60a02ecc3c6d9636232d7371615f4eb306e81 driving_audio.py
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/MiniMax_H3_Continuum_V38X2.json
40199d2500174ba51c040d6983f88d74c4bfbaf19191e728818e6b969b869b0f examples/MiniMax_H3_Continuum_V39.json
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2.json
f3d8dd0f0872afdd97fe24e58b5cb74190d03e3583a66a5d013fee41f6894cda examples/workflows/MiniMax_H3_Continuum_V38X2.zip
705d1141842863b82abe26a8bb015d45eff3bc81ca5bdc53e58b464b743173c0 examples/MiniMax_H3_Continuum_V39_Turbo_Sample.json
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json
4fa8680037595f55f97715f9a2f63d28f11328a87f5c33c316cc4b0ae817e668 examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip
67e81fbdd4c21a8d64fb3bedb5e495698fc8df5cdd9540c8ef5476caa76119c0 examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.json
65689fd800a60ab6939bea020d630616b03a453cf6d48022cd1361992a765abc examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.zip
1b2212b7511fac0b63dfa6e9f006cd8644c72d08c6945ab0e4ed0e722c97a730 examples/workflows/MiniMax_H3_Continuum_V38X2.json
f3d8dd0f0872afdd97fe24e58b5cb74190d03e3583a66a5d013fee41f6894cda examples/workflows/MiniMax_H3_Continuum_V38X2.zip
40199d2500174ba51c040d6983f88d74c4bfbaf19191e728818e6b969b869b0f examples/workflows/MiniMax_H3_Continuum_V39.json
f69b734b5bf19fb6db78bbe4737db0bef69c2a6cef2710f28767882ef35ec847 examples/workflows/MiniMax_H3_Continuum_V39.zip
67e81fbdd4c21a8d64fb3bedb5e495698fc8df5cdd9540c8ef5476caa76119c0 examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.json
65689fd800a60ab6939bea020d630616b03a453cf6d48022cd1361992a765abc examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.zip
705d1141842863b82abe26a8bb015d45eff3bc81ca5bdc53e58b464b743173c0 examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.json
2096e583c6805c0f6f54f76a81e0ed1246325a16d189afba0423fa05e2c19599 examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.zip
a344da581b3c2d77558a460d2b5256e6c9a63b5008c98b52999ceda2e0b2b767 graph_contract.py
bf0556b076dbaca67c3405a7d86c333ca7fdf3d4381529b0a9edc906519943e9 guide_timeline.py
1473c384082700f422f2cf9141e0d012d87d2b43ac8a8a433edc7172e6ca7bf6 hardening.py
33bfd17df1c51c5730ee8007ca3a52e0221d5424b2e86398c50003b9a8f4102c INSTALL_JA.txt
23bddc5a41d2e997f4fbf789c7128e32f4f197f1405b2aef1c76794f57482446 install_windows.bat
12bf572a65d57bbc9a72b459f29691c2057768f4fb7a74f9db87f4775e995bda install_windows.ps1
139546ec1ff459945aa89b50f74b5ee69f62a0e07bbddfcae06936091a409a83 layout_adapter.py
c44abbd789091bbead01250dbd5453e60539d2665c69f162335fdbd2936019c2 LICENSE
d3e213005670e94dee33a240349a4d245cf9a3b6602099f11ed30cf5a32e2c21 media.py
52cec83cfffa8d2504c1f64a9614fb59e56b62c4b1f3106fd3b4fa73870ba6e1 metadata.ini
7ad8ea8859b666cb2771f19795ca8c04ab7aafb5c20cff4e80ad17295429ded3 metadata.ini
d7e613e2cc24110213b477bdd1d390234a41f2af3b1ed006a809066d71d658eb model_patch.py
ecb65a5dbf9e1683c5a3cadd82daaea3a90a2b776f6ebabfbe8b07aa93177422 nodes.py
4decb6e44ee6d172352f223abca4c7a8cba2c8ac6ab0f325b8645f60609f60f1 PACKAGE_INFO_JA.txt
069bf25bd44e7b119c285740adf698bbf30d722a9ca636cd78fe06f026b07d67 PACKAGE_VALIDATION.txt
6ba95639b52219fc37b4f51d52cf6879f75c7418a8d8815b3725b3f201a4f1e6 preview_override_provider_public_v1.json
d4cf474cbd1a98867075c45b53b2ab4d47752bb11611c916bac2a6831ca82c0b preview_override_provider_v1.json
3b7e18bd2ba50fd6c8fe71c58ca79a0bf752f362ed78b7357ce1998efc934b4b preview_override_signature.py
b470819f16fc179654dc1600721a63f0a5a14e777f4c004d574b499da6886ba5 public_api.py
7ff970c9f50192030a302e4967f1747b373dcf5584e823883ec2cc5f8c1b46a2 pyproject.toml
4adbf81a12fa7c99dc748b4275aeefaa098596bee67eca2d118b15d034fcc211 README_JA.md
6b1732f58c2eb300cd1493b7ecdc450ce23910676793d834453abc1aba39bf7c README.md
dc56e02aa94561076bee8a836c268c0e45a8b5248f3b40108bcb740252965ec6 pyproject.toml
caa572ea65dafc86850db99f52433d360ac9c4ab36b854292a3dccd5ba32dc76 reference.py
9539f5cb9fb695841073fa58d9246d5a7410bc372f6cd340378185004dfaf978 reference_audio.py
5cb9bfd6810d3a93a637451980a4cc5c1436e5b84cfd1c7a111cd192e876351b reference_video.py
caa572ea65dafc86850db99f52433d360ac9c4ab36b854292a3dccd5ba32dc76 reference.py
467edb2c188d024c2392aafe7423b3f835e16b10f570da76e2f7dda8a962c1c5 requirements.txt
1845cbf7ba4d3b35afd087a15f5a4d92dc73d4858e61982f8ccf431a836a22b1 run_storage.py
6056de6c0a5322a8a51ba5a114538dca2ef67788cd1afba10fedd483b95f6afa state_io.py
985e7647bd2626af5df68ca702cf6ebb38cc09f3ffe0e239d3e8ef256352a5d8 run_storage.py
8f786989af6339be8cec8f2bf602da89344cf96202d29c980af5c07175eea3d3 spectrum_provider_v029.json
8cdf37b165eb68462dd7accf4a1fd63043f02479f9c377778fb9d4b6feaddcc6 spectrum_provider_v029_py312.json
0efc8d65cb6f59042457ad843fe1b77a4bee45e59d3732f5c83fbc75a31ae77b spectrum_provider_v1.json
71801234f6d2e08b660803194653821ca648e5b9a0d0c3f138bc7ccc6fe9d8c6 spectrum_provider_v1_py312.json
448213ae7670a6e231f833ca1901ca295eb5f6bfb8983ca37703bdb3fdf15f99 spectrum_signature.py
3a359a0467725cc292ccb4c86cf0c3d30bb6a114d01f5345792b09ea3d45441d state.py
6056de6c0a5322a8a51ba5a114538dca2ef67788cd1afba10fedd483b95f6afa state_io.py
660a9df28c7d62ad383cda5514c48c7efcb3576f435af023040116d533a56563 storage_api.py
517662b104ab1507bb927a39f25b8503f092c2ba4c39bc277379ba9d9e31b334 storage_fs.py
daeda14d9f5338835661e6738e074a3897d936cd847fe3e78519efabbcdfd0ef storage_memory.py
53da3ac42f4a66e8adc4560049c7c8e7ada82599be8c61f0573c234342ca36ed storage_settings.py
43ec4023dc06c941e5d7a04db4b8a89fb58d744ceb986f6eb9bf534d4fdcb87b temporal.py
0dbe8a5f71c92f3523942343b40787104c8b96781461ba3ba82934a6bdbe5834 THIRD_PARTY_NOTICES.md
f218e1a93c42467c6057c5c1d04d57cc82f6331b97ec9257e4e2dfb9c831876a timeline_video.py
0bc369ee428b8a5bc8e370a81fbab427262a1fdb7805a78c442c91698d65efba tools/__init__.py
2743597d289d9eb12334d1f5cdd630497e237814ddf48b17f6f68da299c01150 tools/p20_file_backed_tensor_poc.py
@@ -77,14 +104,14 @@ dc2b901789e4b024dbcbc4024c63eb8cbc564d77003ce8c35c0584851ba6dfbd v2/prompts.py
be0d8585fa57e0ccd6d02577bbc7299629e5bfe20e186747f571fccc98b67d7e v2/seam_types.py
12cfcc0c71ca087fe39a5ce2922959a62b2bb311bfe664f05c4480affce03708 v2/seeds.py
0de733c55bb61e69d48918eb5bc91318fc7cf3af383fb8937522ced9df018c47 v2/sequence.py
13213e7709fd02bb810af3c18e4cd7d010f8cdfb121a4b99866a9f2333331089 v2/session_io.py
3ea23dbce6fcd217b6e81235855479b9b16fed7a21693cf156f9014555c4fe13 v2/session.py
13213e7709fd02bb810af3c18e4cd7d010f8cdfb121a4b99866a9f2333331089 v2/session_io.py
78c830cdf49034c6e3b2cd3da5914d5cfc5638eb63ec6dbd1bab13c50d6d4caf v3/__init__.py
ea2922567df90436dd6fa21696833a49dd0cbfeee142a5fedc3a7c303824d473 v3/adaptive_continuity.py
70bec17d0dce1873b9bb4a51b09daf76833722ce89f005a22b029533a367e62b v3/assembly_v35.py
7815230e0a9540bc2edeaf818e940b7a1fdd817fd217ba8a4f9cb21aaf6a308e v3/assembly.py
70bec17d0dce1873b9bb4a51b09daf76833722ce89f005a22b029533a367e62b v3/assembly_v35.py
3f62fdb35e804f42d7cbf0828b4e87e69a2b99246c3f7467fe4fa698f1d803c9 v3/conditioning_bridge_nodes.py
fa7a355e52fbcfe1f53b16fa1c37af5d839215b49412adb41c90129ae4e9d9f7 v3/driving_nodes.py
6403c2bde3fc0b8cddef78567f663fddb48130628f4c9258fcd679c28fd0cf13 v3/driving_nodes.py
147a1d3fc5e09b323a4c3d13d30e432410389bae52b58330462375b9f307b791 v3/easy_audio_nodes.py
64b3e54b42e5156642f9957ccc374fd190d9fc7145637984e702534fb1bf80b4 v3/easy_image_nodes.py
ac859cb1a6e23b8ceae300d818c24fe1346b649a0d663017a9b3ac64aa96fca3 v3/easy_nodes.py
@@ -121,14 +148,14 @@ ae909b43ad17934c85881d3b0bea81e7b45c215eeb4d1f9f6a9ea72faf633ddf v3/reliability
3f03d8659305d2def41939205d21cdc6f35e32943274e87afe539d15b9635f9b v3/review_control.py
be6610fb9427546d9485a17015cfbdc8786715e6a0967f3e4c1dee35d8768385 v3/runtime_coordinator.py
d0a004aa3030c8adaa5ea788f9acc758fcb4a556f203f9a3f63fa4e4b5772063 v3/sampling_engine.py
452022107cea233aa84d5366ec4943ac5aadbd5fcf786113f1d542a641eac8f7 v3/second_pass_nodes.py
97fcb293e0714c3a74b89f08e11c21dce45d25a3010948be1af9dc059d288187 v3/second_pass.py
452022107cea233aa84d5366ec4943ac5aadbd5fcf786113f1d542a641eac8f7 v3/second_pass_nodes.py
6e71e944313339f14e14e11cec682593159ca5542aa5d38095371da781d60224 v3/targeted_refine_sampling.py
5a044bdd70c546e4446e0c26af92aae88049ffa2e0e2d01b52033a83a3152d0e v3/targeted_second_pass.py
26b13614ba0a5fd8b192a106c9503b2f4b6fe7f9268a0fcb0f544979a49cab34 v3/video_adapter_nodes.py
4178dcfa442abdb06ce276e05bd7bcbaa6da28469ba5e57d3da3c663de4e4a9d v3/video_seam_v35.py
2f07ace1462644f0bee553827b3571535de4d0c2d9ef847bd9e0ca73a4eed5cb v3/video_seam.py
c9e03d72565bac7639576474c258ba6fcb0fc73047bd9adb1292bb6ae6db8495 version.py
4178dcfa442abdb06ce276e05bd7bcbaa6da28469ba5e57d3da3c663de4e4a9d v3/video_seam_v35.py
e124ded4bc8ddbf7b432c80d9900e40ee34214327bc6d83bd9fcf95cb55e7169 version.py
b935ad3bb729948bb3890ce967b0ecbef737a5e51ddfc47a0060fc20aa78dcda video_reference_modes.py
685df8e0fa23e3663ab31da1e4470cd19bdaa0dc16362cb606e874d9d6153c2c web/decode_cache_helper.js
fefafd4550511d80bd85ade46a0476f0cf3de13ddb94a908dff6b175f2a0e199 web/easy_bypass_toggle.js
@@ -136,15 +163,9 @@ fefafd4550511d80bd85ade46a0476f0cf3de13ddb94a908dff6b175f2a0e199 web/easy_bypas
9db0aa94f634578a962d5db49e5c70356040a59df968e3a2e531b327a623081f web/easy_load_image.js
35090816ba7470a7255eb85d003b7e79f8a0065982dacf98a18f41c8ebff655d web/easy_load_video.js
0c68d13a1f8beb7431a32b471f8b0612c42fcc4bf72dd97611a025d158008662 web/easy_v38.js
90bef8b7628106fffa95ac1e27e345fa3ccc7203bd2f608497f46b751f9bb850 web/project_id.js
933b5c675294a20ccec263916a1f4453d5b06a530d660ad8e56389d02d7f3023 web/project_id.js
ae59273f163a116ece418957c557c4fd8c3285e3c60814679b9361d2e6e6fa02 web/reference_audio_ui.js
46c9f75c9462bf26300332448f4431cca801f38964ca0a49bc2c50a88a7cc40f web/reference_image_ui.js
beea70f6e25e1366aba5e2689c31733478472f207dc50cad1562cd877152e4f9 web/reference_images_v39.js
bf41bf08dd0504eb753fa5781e6c2b8c1285758ad4bb7309975bb9231d5f8e41 web/video_reference_mode_ui.js
7fd122ff08232e114a6ef32171217ea6324e980eb7257d560c466b8177a3d3ec audio_compat.py
4546ffb0cbaf270c52d1900908e46bfa2d3c0e2d0035ae5694680890e89aedd8 docs/TEMPORARY_STORAGE_JA.md
660a9df28c7d62ad383cda5514c48c7efcb3576f435af023040116d533a56563 storage_api.py
517662b104ab1507bb927a39f25b8503f092c2ba4c39bc277379ba9d9e31b334 storage_fs.py
daeda14d9f5338835661e6738e074a3897d936cd847fe3e78519efabbcdfd0ef storage_memory.py
53da3ac42f4a66e8adc4560049c7c8e7ada82599be8c61f0573c234342ca36ed storage_settings.py
e99c74b78da1aa18be61cb053fa05e5abd106caa5eb5c814cf585786aae1e7f8 web/reference_images_v39.js
42afab2241d39c1b07ff33d8b1370d4b0f729d81e53b4c99f4bc54c6a96b8800 web/storage_settings.js
bf41bf08dd0504eb753fa5781e6c2b8c1285758ad4bb7309975bb9231d5f8e41 web/video_reference_mode_ui.js
+128 -25
View File
@@ -1,35 +1,110 @@
# Public development log
## 2026-10-19 — Second Pass tiled sigma precedence
## 3.9.9 — KJ Preview Override / Review continuation
- Fix `H3ContinuumSecondPassV35` so the tiled path actually forwards the
connected SIGMAS schedule instead of silently rebuilding it from the node's
`upscale_steps` value.
- Give a dedicated `upscale_sigmas` connection priority over the main SIGMAS
connection, preserve widget fallback behavior for the other tiled controls,
and report the selected sigma source for diagnosis.
- Add focused node-level regression coverage for both sigma precedence cases.
Fix new-Run v6 read-only with the separately audited upstream KJ profile, and
show saved reasons when reuse is unavailable. Keep the original local profile,
Sampling v6 guards, old Takes and official workflow bytes.
## 2026-10-15 — CSGlide-style timeline controls
CPU: **2,075 passed / 2 skipped / 2 known expected failures**, plus overlapping
90-case JS coverage. Upstream KJ GPU and normal-ComfyUI browser Continue both
passed: **Chunk 1 preserved, Chunk 2 generated, one reused / one new**; all four
AV outputs fully decoded. Browser checks used the existing local KJ profile.
These are targeted repair checks, not general quality or all-provider acceptance.
- Match CSGlide's shot-timeline interaction: one prompt editor per clip,
coloured shot bands, and boundary sliders that move adjacent shot edges
together.
- Add `@ Add shot` plus a line-start `@` keyboard shortcut for rapid prompt
authoring without changing the emitted Timeline format.
- Make `Default chunk duration` the only duration control; all clip tabs now
share it and legacy per-clip duration values are ignored.
- Ensure the final shot always ends at the full chunk duration; adding a shot
splits the previous final shot instead of leaving a grey tail.
Restart ComfyUI and hard-refresh after updating. On Preview Override, disconnect
optional VAE/Audio VAE and set Tiny VAE `none`; retain generation VAEs. Correct
unsupported settings before starting a new Run. Old unsafe/v5 Takes remain
readable and are not automatically made reusable. No workflow rewiring or Take
deletion is required. [Full release notes](docs/RELEASE_NOTES_V399.md).
## 2026-10-15 — Sequence prompt builder
- Add `DumasH3SequencePromptBuilder` and its tabbed ComfyUI editor.
- Validate clip-local slot ranges and deterministically convert them to
absolute `[start-end]` timeline blocks, preserving one STRING output for
direct wiring into Continuum.
- Add focused unit coverage for formatting, normalization, validation, and
registration/schema behavior.
## 2026-10-16 — V3.9 Turbo Sample
Add the user-supplied sample separately: Auto, 2 x 8s, multiline image-anchored
dance prompt, fixed background/camera, Thai rap and one enabled Turbo LoRA.
Preserve the supplied 6-step/Draft settings and document their unverified pairing
with the 4-step/768p LoRA. Fresh workflow/project IDs avoid user Run identity reuse;
remove misleading auto-download metadata for the saved custom model. Standard
workflows, existing Takes and runtime code stay unchanged. No new GPU validation,
runtime deployment, release or Registry publication is included.
Local CPU checks: 17 passed, zero CUDA initialization attempts. Workflow link
consistency, gallery/ZIP bytes, saved settings and both manifests were checked.
## 2026-10-16 — Japanese README organization
Align the Japanese README with the compact English entry point, including exact
model/VAE/text-encoder downloads, model folders, external node links and optional
Turbo guidance. Keep current compatibility and known limitations explicit.
Move all former Japanese detail and ten image references into USER_GUIDE_JA.md
and README_HISTORY_JA.md, preserving historical context and rebasing links.
Include these text guides in the Registry selection and update both manifests.
English documentation, product code, images, official workflows and existing
Runs/Takes remain unchanged. This is local documentation work, not a new release.
Local checks: 5 existing tests passed, 95 relative file/heading links resolved,
all ten original image references and original detailed text were retained, and
both manifests matched (464 source entries / 162 Registry entries). The Japanese
README is 247 lines instead of 801. Its 18 external destinations match English.
No CUDA initialization, runtime deployment, commit or push was performed.
## 2026-10-16 — condensed README and retained reference material
Organize the English README as a short installation and operation entry point.
Keep the version heading, fixed-version downloads, getting-started video, Review /
regeneration steps, history safety and bounded v3.9.8 compatibility/known limits.
Move the former detailed controls and screenshots into docs/USER_GUIDE.md, and
release/legacy implementation material into docs/README_HISTORY.md; retain all
original sections and image assets, with rebased links and historical context.
Follow-up: list the template-selected H3 diffusion model, text encoder, Video/Audio
VAEs with provider downloads and ComfyUI model folders. Link Spectrum, KJNodes,
rgthree and optional task-matched LightX2V Turbo LoRAs. Explain the saved INT8
Video VAE versus old FP16 download metadata without altering workflow bytes.
Include the two text references in the Registry selection and update both source
manifests. Product code, Japanese README, official workflows, stored Takes,
release tags/assets and runtime deployment are unchanged by this documentation work.
Local documentation checks: 5 existing tests passed; all 143 relative file/heading
links resolved, all 39 original image references were retained, and both manifests
matched (462 source entries / 160 Registry entries). README is 267 lines, down
from 1,574. No code, workflow, image or Japanese README changes were made.
## 2026-10-16 — published v3.9.8 GPU recheck documentation
Record a fresh GPU recheck after deploying and restarting the exact published tag `58fbd855`: PyTorch/KJ and Comfy Kitchen INT8 + audited local Spectrum/KJ each passed separate-Queue Chunk 1 to 2 continuation. KJ retained images, contracts and original Chunk 1 stayed equal, and all four AV outputs decoded. Existing Takes/workflows were hash-protected. README and Release notes distinguish this published-tag result from the earlier candidate gate and retain exact provider/no-preview-VAE/Tiny-none scope plus browser, quality and performance limits. This is a documentation-only main update; the original v3.9.8 tag and package assets remain unchanged.
## 2026-10-16 — package 3.9.8 / KJ Preview Override Phase A
- Audit the actual KJ Preview Override file, SHA-256 `d07c8e5ff7d7abd3e31acefe2cf29f618b5da5341b3b434e72ec2ac0953230b0`, rather than accepting KJNodes 1.5.2 by version alone. Pin sibling helpers and Core preview/dispatcher source, then compare loaded code/defaults/classes without importing or executing the provider during observation.
- Add a dedicated adapter only at `outer_sample/kj_preview_override`. Preserve all fixed wrapper settings and Core's key/list order. Ignore only the exact builtin transient `frames` list after the provider audit; preserve the generic no-KJ observer and all other reuse guards.
- CPU scope covers real-provider frame changes, reconstruction, fixed-setting/order changes, monkeypatch/source rejection, VAE/Tiny VAE exclusion, combined Attention/Spectrum identities, saved prefix bytes and old unsafe history. Full real-Core CPU gate: 1,995 passed / 3 skipped / 2 existing Inspector expected failures / zero unexpected failures and CUDA initialization attempts; the 48 dedicated KJ tests passed without skips. Normal ComfyUI GPU acceptance passed four Core-API Queues: PyTorch and Comfy Kitchen INT8 + the audited local Spectrum implementation, each with separate-Queue Chunk 1 to 2 continuation and held preview images. Both second Queues reused one chunk and generated one; original Chunk 1 bytes and contracts were unchanged. All four video/audio outputs fully decoded. See docs/KJ_PREVIEW_OVERRIDE_V398.md for exact support and limits. GitHub main/tag CI, Release asset integrity and Registry status are separate publication checks.
## 2026-10-15 — package 3.9.7 / linked-input regeneration cache
- Keep Full Run automatic variations reachable when Core leaves linked Chunks, Nonce, storage, regeneration boundary or generation mode unresolved during cache-key evaluation.
- Validate the actual chunk range at execution, and retain stable caching for known ordinary Full Run, storage OFF and directly specified fixed nonces. An unresolved external nonce conservatively reevaluates the node; Run Storage still owns prefix reuse.
- Keep Sampling Contract v6, Take protection, public schemas and official workflows unchanged. Add direct-input and real-Core cache regressions for both V3.8 and V3.9.
## 2026-10-15 — package 3.9.6 / regeneration intent and automatic variations
- Separate saved manual regeneration intent from the legacy stale Chunk 1 / nonce 0 state. Retain earlier unmarked later-chunk/fixed-nonce settings, and clear intent metadata when Auto or a one-shot restart consumes it. No widget order or API inputs change.
- Bypass Core output caching only for Full Run explicit regeneration with Run Storage on and automatic nonce 0. Run Storage remains the authority for compatible prefixes, interrupted variations and new completed variations; its Sampling v6 checks are unchanged.
- Add restored-workflow Queue cases for both public Samplers and a CPU-only real-Core cache lookup gate. Keep historical unsafe/v5 Takes read-only and official workflow bytes unchanged. Full CPU gate: 1,908 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and CUDA initialization attempts. Normal ComfyUI browser save/reload retained Chunk 2 / nonce 5 and Chunk 1 / nonce 0 after history loading. Five new-Run GPU queues passed continuation, Chunk 2 regeneration and two identical Full Run / nonce 0 submissions; earlier Chunk 1 bytes were preserved. GPU scope: PyTorch, Spectrum OFF, 512x512, 8-step Turbo; all five media outputs fully decoded. See docs/REGENERATE_INTENT_V396.md. GitHub main/tag CI, Release asset integrity and Registry status are verified separately during publication.
## 2026-10-15 — package 3.9.5 / Issue #31 publication
Main Linux/Windows CI passed. Correct the tag-presence probe to branch on the API exit status: a missing-ref error body must not be treated as an existing commit SHA. Product code and accepted GPU signatures are unchanged.
Consolidate the Attention fix and finite Spectrum binding adapter on the exact 3.9.4 public source tree. Add a separate audited upstream Spectrum 0.2.29 profile without changing existing local-provider identities. Keep old unsafe history, public node IDs, contracts and all official workflows. CPU/GPU and publication checks are recorded separately in docs/ISSUE31_SAMPLING_SIGNATURES.md.
## 2026-10-14 — Issue #31 Phase 1, unpublished
- Final standard CPU validation: 1,777 passed / 1 skipped / 2 existing Inspector expected failures; no failures. All three standard commands recorded zero CUDA initialization attempts. Native PackedLayout, prompt planning and twelve-node registration checks passed.
- Limit production edits to the MODEL attention signature adapter in run_storage.py. Use already-loaded Core registries and implementation bindings; never import Core/Kitchen or execute an attention kernel during observation.
- Include both container functions and provider implementations/state without persistent addresses. Keep unrelated v6 signatures and every existing reuse guard; do not rewrite old unsafe manifests or raw files.
- Add real-Core CPU signature, clone/recreation, separate-process/hash-seed, mutation detection, prefix reuse and history immutability regressions. Kernel registration in CPU fixtures is structural evidence, not INT8 execution acceptance.
- GPU Review continuity, Spectrum/KJ execution, browser behavior and reporter-exact reproduction remain separate. No public node, official workflow, installed runtime, package version, commit or release changes.
## 2026-10-14 — package 3.9.4 Registry packaging repair
@@ -84,3 +159,31 @@ Earlier product behavior is described in CHANGELOG.md and the historical Release
- Add 3.9.3 and retained 3.9.2 Release, custom-node ZIP and checksum links to the current English/Japanese guides. Update the installation heading to 3.9.3 and retain the implemented settings screenshot and historical validation records.
- Refresh Registry then development manifests from saved bytes. The existing release-metadata/public-surface CPU gate passes 8 tests with zero CUDA initialization attempts; all 25 tracked JSON/ZIP assets retain their bytes.
- Product code, package metadata, public schemas and saved-data contracts are unchanged. GitHub Release/asset and Registry publication are verified separately; no new GPU, long-form performance or physical SSD-write validation is claimed.
## 2026-10-15 — Issue #31 Spectrum binding identity candidate
Added a dedicated, fail-closed Spectrum binding adapter without changing the generic observer, Sampling v6, runtime signature v1, Graph v4, public nodes, workflows or package version. The immutable config fields and audited source/loaded-callable contract identify generation; reviewed runtime history, counters and profile caches do not change persistent identity. Unsupported structures, additional attributes and altered implementations remain non-reusable. Existing unsafe Takes are preserved without migration. CPU and GPU evidence are recorded separately; no commit or publication is authorized.
Issue #31 Phase 3 validation: the final standard CPU entry point passed 1,834 tests, 1 skip and 2 existing expected Inspector failures, with zero CUDA initialization attempts. The 57 new Spectrum tests cover all 27 config fields, transient-state/process stability, implementation/extra-state refusal, unaffected v6 contract identity and old unsafe Take protection; the 36 Phase 1 attention tests also pass. Under one audited local provider implementation, both PyTorch+Spectrum and Kitchen+Spectrum persisted new Chunk 1 Takes and reused them from a separate Queue to generate Chunk 2, with unchanged prefix hashes, identical safe contracts and fully decodable 10-second AV output. This is a targeted local GPU gate, not acceptance of other providers, KJ, browser behavior, subjective quality or benchmarks. No commit, push or publication occurred.
## 2026-10-20 — External Chunks Reference assignment table candidate
Only the V3.9 Reference Images helper presentation changes: external Chunks links make the destination count unresolved, expose assignment range 1–16, and label All/Off with that range. Direct counts retain the existing table. The frontend does not resolve arbitrary upstream values or alter native Chunks, saved selectors, Sampling identity or stored Takes. Added 21 JavaScript cases and four public-input CPU cases cover stale widget counts, zero/unrestored links, multiple Samplers, range actions, future assignments and callback stability. Existing CPU coverage passed 310 tests with two known Inspector expected failures; the four added CPU cases and metadata checks also pass. Targeted browser acceptance passed for external INT with stale widget values, live disconnection/restoration, actual KJ Set/Get routing, all-page All/Off actions, saving/reloading and workflow tab switching. The candidate display code was deployed to the chosen runtime and reloaded without restarting its backend. Existing saved Takes and workflows were preserved. No GPU generation, commit, push or public release occurred.
## 2026-10-20 — Spectrum source/runtime audit portability candidate
Normalize CRLF to LF in Spectrum provider-file and module-closure source checks. Add fixed Python 3.12 profiles generated from the same hash-verified 44-file and upstream 0.2.29 55-file sources. Retain the original Python 3.13 profile bytes and valid identity descriptors. Unknown code, file sets, closures and runtime extensions remain non-reusable; unsupported Python runtimes are not promoted. Old unsafe/v5 Takes remain readable without conversion, and public nodes, Sampling v6, Graph v4 and official workflows are unchanged.
The full standard CPU/JS gate passed 2,096 tests with two skips and two existing Inspector expected failures; compile and offline runtime verification passed with twelve public nodes. Targeted Spectrum/Attention coverage passed 169 tests with one legacy-profile skip. Python 3.12 component coverage passed 112 tests with one legacy-profile skip, checking all 27 configuration fields, code/state refusal and newline parity; 21 independent older-Core/Attention cases were outside that component gate. Separate-process storage checks passed 16 PyTorch/Kitchen cases on Python 3.13 and eight Spectrum/default-model cases on Python 3.12. Actual Take read-only JavaScript agreed with all 24 storage results, and all six previously valid complete Sampling Contracts remained identical. No CUDA initialization, new GPU generation, browser acceptance, commit, push or publication is claimed. The older Core/Kitchen observation limitation remains separate from Spectrum recognition.
## 2026-10-20 — 3.9.10 publication preparation
Prepare the tested Spectrum audit portability and Reference assignment display
repairs as the next 3.9 patch. Use SemVer 3.9.10 for GitHub and Registry while
reserving 4.0 for the major update. Preserve existing 3.9.9 and historical tags,
assets and official workflow bytes. Publish CPU/browser scope explicitly without
claiming new GPU acceptance. Version metadata, release workflows and package
manifests identify the same candidate; public asset and CI checks follow upload.
+66
View File
@@ -0,0 +1,66 @@
# Issue #31 — Attention and Spectrum reuse signatures (3.9.5)
## 3.9.10 — Spectrum audit portability
The repair treats source CRLF and LF line endings as equivalent
in both the provider-file audit and provider module closures. All other source
bytes and the exact supported Python file sets remain checked. The canonical
source hashes are unchanged, so existing valid Python 3.13 signatures retain
their identities.
Separate `spectrum_provider_v1_py312.json` and
`spectrum_provider_v029_py312.json` profiles identify the same audited source sets
under Python 3.12. They were generated and checked with Python 3.12.10; the
original profiles remain byte-identical for Python 3.13.12. Other Python runtimes,
altered implementations, foreign closures and unknown runtime fields remain
non-reusable. The adapter does not execute provider functions to approve reuse.
This repairs Spectrum observation, not every Core/Kitchen/KJ combination.
Independent Attention/provider checks must also pass. In particular, an older
Core/Kitchen environment can still fail its own Attention observation even when
Spectrum is recognized. Do not upgrade Python or replace Attention solely to
make an old Take reusable. Keep old history and start a new Run after updating.
No new GPU or reporter-exact acceptance is claimed for this repair.
The standard CPU/JS gate passed **2,096 tests / 2 skips / 2 known expected
Inspector failures**, with compile and offline registration checks passing.
Separate-process storage probes passed 16 Python 3.13 PyTorch/Kitchen cases and
eight Python 3.12 Spectrum/default-model cases; actual Take read-only JavaScript
matched all 24. The six previously valid full Sampling Contracts stayed exact.
These CPU checks execute no Attention kernel or H3 checkpoint inference.
### 日本語 — 3.9.10の局所修正
Spectrum監査でCRLF/LFの改行差を同一として扱い、対応済みソースにPython 3.12用の
固定プロファイルを追加します。既存3.13の正常な識別値と旧Takeは維持します。
未知のコード・追加状態の拒否は継続します。Spectrumが識別可能でも、Core/Kitchenなど
別の観測に失敗する環境は再利用不可のままです。適用後は新規Runで確認してください。
旧履歴を利用可能にする目的でPythonやAttentionを変更する必要はありません。
## 3.9.8 Preview Override Phase A (2026-10-16)
The new adapter is restricted to the KJ source profile in `preview_override_provider_v1.json`: Preview Override SHA-256 `d07c8e5ff7d7abd3e31acefe2cf29f618b5da5341b3b434e72ec2ac0953230b0`, audited sibling helpers and Core preview/dispatcher files. The complete profile digest is part of the persistent identity, so later dependency-profile upgrades cannot retain the same identity accidentally. It compares loaded functions, nested code, defaults and supported classes with the pinned source; it never imports KJNodes, executes a preview callback, loads a VAE or initializes CUDA during signature observation. Source line endings are normalized to LF for provider auditing.
After that audit, the exact builtin `frames` list is treated as execution-local display data. Max resolution, node ID, JPEG quality, suppression, frame count, FPS and supported Core preview settings remain part of the identity. Core's cross-key insertion order and each wrapper list's order are preserved explicitly, because the generic dictionary observer sorts keys. The ordinary non-KJ signature path is unchanged.
Phase A requires `vae is None`, `audio_vae is None`, KJ Tiny VAE `none`, and Core NoPreviews/Auto/Latent2RGB. Unknown files, replaced code/defaults/helpers, extra wrapper state, foreign classes and approximate-VAE previews keep reuse disabled; they are not promoted by an empty frame list. This profile does not claim support for every KJNodes 1.5.2 checkout or the different upstream Preview Override source. No general `frames` exclusion is added.
Existing normal non-KJ Sampling v6 identities remain exact. Earlier safe KJ identities may differ and cannot be assumed reusable; earlier unsafe/v5 histories remain preserved and read-only. Start a new Run after updating; old KJ history is not migrated. Full real-Core CPU gate: 1,995 passed / 3 skipped / 2 existing Inspector expected failures / zero unexpected failures and CUDA initialization attempts; the 48 dedicated KJ tests passed without skips. Normal ComfyUI GPU acceptance passed four Core-API Queues: PyTorch and Comfy Kitchen INT8 + the audited local Spectrum implementation, each with separate-Queue Chunk 1 to 2 continuation and held preview images. Both second Queues reused one chunk and generated one; original Chunk 1 bytes and contracts were unchanged. All four video/audio outputs fully decoded. See docs/KJ_PREVIEW_OVERRIDE_V398.md for exact support and limits. Product files used in this targeted GPU gate are byte-identical to the 3.9.8 repair. Browser Run-button behavior, subjective preview quality, reporter-exact reproduction and per-kernel return tracing are not newly accepted.
## Previous 3.9.5 Attention / Spectrum repair
With Model Attention Backend between a LoRA Loader and Spectrum, cyclic Attention closures and a slots-based SpectrumH3Binding could be unobservable. Generation still succeeded, but Sampling v6 correctly refused reuse with "Saved history is read-only". This update identifies the reviewed runtime implementations rather than forcing resume_safe.
Core PyTorch and Comfy Kitchen INT8 overrides are identified by their loaded functions, defaults, closures, both container paths and numerical/provider state. Spectrum identity includes all 27 effective immutable config fields and audited source/loaded-function contracts. Reviewed history, counters and statistics do not change persistent identity. Unknown sources, foreign module closures, extra runtime state and altered implementations remain non-reusable.
Two explicit Spectrum source profiles are included: upstream v0.2.29 at commit 806fe498d574a726fecb8818917f00a4b993fd77 (55 provider files), and the prior audited 44-file implementation. Versions/class names alone do not enable reuse. The loaded-function contract is environment-specific: validation uses Python 3.13.12, ComfyUI Core 0.40.0 and comfy-kitchen 0.2.37; other source/runtime combinations may remain read-only. BSA/Flow/Untwist and KJ/Sage patch-node combinations are outside this targeted acceptance.
Update the custom node and restart ComfyUI. Start a new Run. Existing unsafe/v5 Runs are preserved and are not repaired or promoted; normal unaffected Sampling v6 identities remain compatible. Sampling v6, Graph v4, Run Storage v3, public IDs/socket order and shipped workflow bytes stay unchanged.
The signature gate passed 152 CPU tests with one original-profile-only skip and no failures; CUDA initialization attempts were zero. Full standard validation passed 1,893 tests with 2 skips and 2 existing Inspector expected failures (zero unexpected failures and CUDA initialization attempts). Public upstream 0.2.29 PyTorch/Spectrum and Kitchen/Spectrum new-Run GPU gates both passed separate-Queue Chunk 1→2 continuation, prefix/contract preservation and complete 512x512 / 24fps / 10-second AV decoding. Actual CUDA backend returns were observed. Existing Inspector reconfigure expected failures, browser acceptance, visual/audio quality and benchmarks remain separate.
## 日本語
Model Attention BackendとSpectrumを併用したReview継続で、Attentionの循環参照とslots付きBindingを観測できず、生成後のTakeが再利用不可になる問題へ対応します。全27設定値・実ソース・読み込まれた関数を識別し、実行中の履歴や統計値は永続署名へ含めません。resume_safeを強制せず、未知の実装や追加状態は安全側で拒否します。
公式Spectrum v0.2.29の固定コミットと、以前に監査した44ファイルの実装を別プロファイルとして扱います。別のSpectrum版、別のPython/Core実装、BSA/Flow/UntwistやKJパッチの組合せ全般を保証するものではありません。更新・再起動後は新しいRunを作成してください。以前のunsafe履歴を自動修復せず、既存TakeとWorkflowは保持します。
+95
View File
@@ -0,0 +1,95 @@
# KJ Model Preview Override continuation — 3.9.8 / Phase A
The audited Preview Override wrapper can keep its transient image list without changing the MODEL sampling identity. The adapter retains fixed settings, source and loaded implementation checks, the complete audit-profile digest, and Core's cross-key/list wrapper execution order. It does not add a generic exception for any field named `frames`.
## 3.9.9 public-provider repair
The source checkout additionally accepts the pinned upstream implementation from
[KJNodes commit 17195fc](https://github.com/kijai/ComfyUI-KJNodes/blob/17195fcfa8a05f601b766a0f09c9733284f47416/nodes/preview_override_node.py).
Its LF-normalized Preview Override SHA-256 is
`1916295463eb790d3e8ea2b0dae2b25344549479b92e90a1d238acddb2dde3ab`,
profile `kj-preview-upstream-17195fc-phase-a` in
`preview_override_provider_public_v1.json` (canonical profile SHA-256
`da128a4324d8e0a63d893501f8d65652d25da60690afaf1447d80f2a2b67617f`). Its sibling and Core hashes are the
same as the original table below. The wrapper implementation is unchanged; the
node factory additionally resolves display IDs through Core's DynamicPrompt.
Both source and loaded-code audits remain required. This does not cover arbitrary
KJNodes versions, future Core changes or preview-VAE configurations.
The original profile JSON and successful descriptors are preserved. Profile
registration alone cannot invalidate another profile's Takes. KJ-free contracts
are unchanged. Missing or malformed optional public-profile data disables only
that profile. Ordinary generation keeps its existing fallback.
Review and Render History distinguish v5 history from v6 Takes whose generation
conditions could not be verified. Bounded saved reasons identify unsupported
preview settings or implementations; missing metadata points to Detailed Report.
Reasons come from the selected Take's storage revision and do not affect its
sampling identity or enable disabled actions.
For an unsupported KJ/Core implementation, bypass **Model Preview Override** in a
workflow copy and test a new Run. For unsupported preview VAE settings, disconnect
only that preview node's optional VAE/Audio VAE and use Tiny VAE `none`; keep the
generation VAEs connected. For Core approximate-VAE preview, use Auto or
Latent2RGB. Correct the configuration before starting a new Run: repeatedly
restarting Chunk 1 with the same unsupported setup will not repair it.
Old unsafe/v5 Takes remain readable and are never promoted or rewritten.
Full real-Core CPU regression: **2,075 passed, 2 skipped, 2 known Inspector
expected failures, zero unexpected failures**. The 90-case production-JS fixture
also passed with eight real Run Storage project vectors. These checks overlap
the pytest total and must not be added together. Unmodified 3.9.8 fixtures confirm
exact legacy/KJ-free identities and byte-identical saved Takes; all 21 existing
workflow JSON/ZIP files are unchanged. Offline registration retains twelve nodes
and passes native PackedLayout / Fixed 3x5s checks. Recorded CPU drivers attempted
zero CUDA initializations. The skips cover a local-only Spectrum descriptor
fixture on upstream Spectrum and an unavailable historical V3.5 workflow.
The additional public profile passed two targeted H3 GPU Queues with cached
preview images and unchanged Chunk 1. Normal-ComfyUI browser Continue passed
separately with the retained original local profile. See
[3.9.9 validation and limits](RELEASE_NOTES_V399.md). The historical v3.9.8 tag
and assets retain their original local-only profile.
## Exact supported profile
The published v3.9.8 package accepts only the following explicitly audited source combination. KJNodes reports 1.5.2, but that version alone does not identify the supported implementation: this Preview Override file differs from upstream. The 3.9.9 extension is described above. Files are LF-normalized before source hashing. An unsupported provider remains non-reusable rather than blocking ordinary generation.
| Source | SHA-256 |
| --- | --- |
| KJ `nodes/preview_override_node.py` | `d07c8e5ff7d7abd3e31acefe2cf29f618b5da5341b3b434e72ec2ac0953230b0` |
| KJ `nodes/tiny_vae.py` | `61bf4daa634ef6bb901ff8630071caed12e733d17e7c4ba50082dc0aa5506591` |
| KJ `nodes/ltxv_nodes.py` | `11286583c56e653dbd57f5e8eda6902c84b759f16b359b3bf0c895b43e6b5364` |
| Core `latent_preview.py` | `b323be881ade3388254a72f5463e329f6d0a263b0078b69ae4256eb4e22dbba2` |
| Core `comfy/patcher_extension.py` | `1c7b3b855278d0b95d1c1a705fa0b3c91a265f320e2f97bdeb9d15469d6d4ac7` |
Profile: `kj-preview-1.5.2-local-d07c8e5-phase-a`. Canonical full-profile SHA-256: `d146bd89895fef5a8d576134d519e63bbf4ac1c35188f70b4de4e4ebfb498fc7`. The provider JSON is shipped with the runtime adapter.
Preview Override's optional VAE and Audio VAE must be disconnected, Tiny VAE must be exactly `none`, and Core preview must be NoPreviews, Auto or Latent2RGB. A generation Video/Audio VAE is still used normally; the restriction concerns the preview node's optional connections. Core approximate-VAE/TAESD, altered helpers, unknown wrappers, extra state and other providers are outside this profile. Source alone is insufficient: loaded functions, nested code, defaults, classes and selected bindings must match the audited implementation.
## Update and history protection
Restart ComfyUI, refresh the browser and create a new Run. Do not delete or rewrite old Runs/Takes. Earlier safe KJ identities may differ; old unsafe/v5 history is not repaired or promoted. Ordinary non-KJ Sampling v6 signatures remain exact. Sampling v6, Graph v4, Run Storage v3, State/Session formats, public node inputs and official workflow files are unchanged.
## Validation
Full real-Core CPU validation: 1,995 passed, 3 skipped, 2 existing Inspector expected failures, zero unexpected failures or CUDA initialization attempts. The dedicated real-provider tests passed all 48 cases, including 0/1/multiple held frames, reconstruction, fixed-setting changes, wrapper ordering, source/code/default/helper changes, VAE exclusion, Attention/Spectrum combinations, prefix preservation and old unsafe history. Offline registration retains twelve nodes.
The initial acceptance used the same repair product bytes. A subsequent recheck installed all 464 files from the published v3.9.8 tag (`58fbd855`), restarted normal ComfyUI, verified the live 3.9.8 startup banner and saved adapter/profile identity, and passed four new-Run GPU Queues:
| New Run | First Queue | Separate second Queue |
| --- | --- | --- |
| Model Attention Backend PyTorch + KJ Preview Override | Generate Chunk 1; 8 held images | Reuse Chunk 1, generate Chunk 2; 16 held images |
| Model Attention Backend Comfy Kitchen INT8 + audited local Spectrum + KJ Preview Override | Generate Chunk 1; 8 held images | Reuse Chunk 1, generate Chunk 2; 16 held images |
In both second Queues the KJ MODEL node was cached, the Sampler executed, the stored contract/revision stayed equal, and original Chunk 1 latent bytes were unchanged. All four video/audio outputs fully decoded (first Queue 5 seconds / 120 frames; second Queue 10 seconds / 240 frames, 512x512, 24fps). Saved contracts recorded Sampling v6, resume_safe=true and the pinned adapter/profile identity. Runtime source hashes and stored descriptors verified the repair was actually loaded.
Conditions: RTX 5060 Ti 16 GB, Python 3.13.12, PyTorch 2.14.1+cu130, Core 0.41.0, frontend 1.55.17, comfy-kitchen 0.2.37; 8-step Turbo LoRA, Euler/simple, 2x5-second chunks, Review Each Chunk / Continue, saving and Auto Resume on. Preview Override: resolution 512, quality 80, one frame, 12fps, suppress default on, VAE/Audio VAE absent and Tiny VAE none; Core Auto. Spectrum used the previously audited local implementation with offline smoothing/replay, system-RAM history/archive and model-aware off. This does not newly validate upstream Spectrum 0.2.29 combined with KJ, other wrapper orders, Sage patch nodes, or all preview configurations.
These were controlled submissions through the normal Core API. Browser Run-button/save-reload behavior, reporter-exact reproduction, per-kernel CUDA return tracing, subjective audiovisual/preview quality and performance benchmarks are not newly accepted. Held PNGs were 32x32, so the image counts do not establish high-resolution preview quality. The two known Inspector reconfigure expected failures remain. Existing workflow files and prior Takes were hash-protected; no history migration occurred. GitHub Release/package verification, Registry upload/moderation and real Manager installation remain distinct publication checks.
## 日本語
監査済みKJ Preview Overrideの一時画像だけを永続署名から分離し、固定設定・実装識別・ラッパー順序を保持します。KJNodes 1.5.2全体ではなく、上表のソースと読み込み済み関数を照合できる実装だけが対象です。VAE/Audio VAE未接続・Tiny VAE none・Core NoPreviews/Auto/Latent2RGBが条件です。
更新後は再起動し、新Runを使用してください。旧Take・Workflowの削除や移行は不要で、旧unsafe履歴は再利用不可のまま保持します。CPU 1,995成功(専用48件を含む)、PyTorchおよびKitchen+ローカルSpectrumの各2 Queueで画像保持後のChunk 1→2継続、元Chunk 1保持、4本の映像・音声全デコードを確認しました。初期の同一製品コードによる受入後、公開タグv3.9.8(58fbd855)を実環境へ反映・再起動して、新Runで同じ2構成の4 Queueも再確認済みです。画質・速度・実ブラウザー操作・全KJ構成への合格は主張しません。
+905
View File
@@ -0,0 +1,905 @@
# H3 Continuum — README history and implementation reference
[Current README](../README.md) · [Illustrated user guide](USER_GUIDE.md) · [CHANGELOG](../CHANGELOG.md)
This reference preserves the release descriptions, earlier screenshots, detailed
validation results and pre-V3.8 implementation notes previously shown in README.
It was extracted from source commit `894869d9` (package 3.9.8).
Statements such as "current on main", historical installation commands, node lists
and old acceptance results retain their original context; they are not current
installation recommendations or broader validation claims.
For current downloads, compatibility and known limits, start with the README.
- [v3.9.8 and published-tag GPU conditions](#398--kj-model-preview-override-continuation)
- [Older implementation notes](#archived-implementation-notes-pre-v38-not-the-current-guide)
- [Prompt formats and examples](#prompt-formats)
- [Historical resource and validation limits](#limits)
<!-- Preserved README sections begin. -->
**Latest package: 3.9.8.** Use the fixed-version GitHub downloads below. Registry publication is verified separately.
| Version | Release | Custom-node ZIP | SHA-256 checksum |
| --- | --- | --- | --- |
| 3.9.8 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.8) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/ComfyUI-H3-Continuum-v3.9.8.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/SHA256SUMS.txt) |
| 3.9.7 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.7) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/ComfyUI-H3-Continuum-v3.9.7.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/SHA256SUMS.txt) |
| 3.9.6 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.6) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/ComfyUI-H3-Continuum-v3.9.6.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/SHA256SUMS.txt) |
| 3.9.5 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.5) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/ComfyUI-H3-Continuum-v3.9.5.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/SHA256SUMS.txt) |
| 3.9.4 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.4) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/ComfyUI-H3-Continuum-v3.9.4.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/SHA256SUMS.txt) |
| 3.9.3 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.3) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/ComfyUI-H3-Continuum-v3.9.3.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/SHA256SUMS.txt) |
| 3.9.2 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.2) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/ComfyUI-H3-Continuum-v3.9.2.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/SHA256SUMS.txt) |
## 3.9.8 — KJ Model Preview Override continuation
An audited KJ Model Preview Override can keep preview images without changing a new Run's sampling identity. Separate Queues can reuse Chunk 1 and generate Chunk 2 while preserving earlier Takes. Support is limited to the explicitly pinned KJ source and Core helpers, **VAE / Audio VAE disconnected, Tiny VAE `none`**, and Core NoPreviews/Auto/Latent2RGB. Other KJ builds remain conservative; version 1.5.2 alone is insufficient. Restart ComfyUI, refresh the browser and start a new Run; do not delete old Takes or workflows. See [compatibility, source hashes and verified scope](KJ_PREVIEW_OVERRIDE_V398.md).
### Published v3.9.8 GPU recheck — 2026-10-16: PASS
After restarting the normal ComfyUI installation with the published tag **`v3.9.8` / `58fbd855`**, all 464 installed release files matched the tag. The startup log confirmed 3.9.8, and new saved Sampling v6 contracts contained the KJ adapter and pinned audit-profile digest. Two brand-new Runs passed four Core-API GPU Queues:
| KJ Preview Override configuration | Separate-Queue Chunk 1 → 2 |
| --- | --- |
| Model Attention Backend: PyTorch, Spectrum OFF | **PASS — 1 reused / 1 generated** |
| Model Attention Backend: Comfy Kitchen INT8 + audited local Spectrum | **PASS — 1 reused / 1 generated** |
On both second Queues, the KJ MODEL stayed cached with preview images retained (**8 → 16**), while the Sampler executed. The contract/revision and original Chunk 1 latent hashes stayed unchanged; `resume_safe=true`. All four video/audio outputs fully decoded (5 seconds / 120 frames, then 10 seconds / 240 frames, 512×512 at 24fps). Existing workflows and Takes were preserved: 170 protected files, including 71 saved workflow files and 46 pre-existing Run/Take files, retained their hashes.
**Scope:** only the exact audited KJ/Core profile; the local KJ source differs from upstream, so KJNodes 1.5.2 alone is insufficient. Preview Override VAE / Audio VAE were disconnected, Tiny VAE was `none`, Core preview was Auto. GPU: RTX 5060 Ti 16 GB, 8-step Turbo, Euler/simple, 2×5-second chunks. Spectrum used the audited local offline-replay implementation, system-RAM history/archive and model-aware OFF. **Unverified:** browser Run-button/save-reload behavior, the reporter's exact complete graph, other KJ/Spectrum builds, SolAttn/preview VAE paths, per-kernel return tracing, subjective quality and speed/VRAM benchmarks. Retained PNGs were 32×32, not high-resolution preview-quality evidence. The known Inspector issue remains; old unsafe/v5 Takes are not repaired or promoted. [Issue #31 result](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/31#issuecomment-6209571261).
## 3.9.7 — External-input regeneration cache fix
Fix automatic Full Run variations when Chunks or Variation Nonce come from external nodes. Unresolved linked inputs no longer let Core skip an explicitly requested regeneration. Earlier accepted chunks, existing Takes, public inputs and official workflows are preserved. Update the custom nodes, restart ComfyUI and refresh the browser; no workflow changes or history deletion are required. [Release notes](RELEASE_NOTES_V397.md).
**Published v3.9.7 GPU recheck (2026-10-15): PASS.** After restarting with the published tag (`18df80d`), external Core INT nodes supplied Chunks=2 and Nonce=0. Two completely identical Chunk 2 regeneration prompts advanced automatic Nonce **1 → 2**; both reused Chunk 1 and generated only Chunk 2. Earlier Takes and raw latents were retained. All three outputs (baseline plus two variations) fully decoded with 240 frames and 10-second video/audio. Existing 60 workflows, one settings file, 30 Run/Take files and all 459 release files remained hash-identical. Scope: RTX 5060 Ti 16 GB, 512×512, 24 fps, 2×5 seconds, 8-step Turbo, PyTorch attention, Spectrum OFF, Core API submissions. No CUDA/OOM/execution errors were observed; this does not extend browser, subjective-quality or other-accelerator acceptance.
The original [3.9.2 GitHub source archive](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/archive/refs/tags/v3.9.2.zip) remains available from its unchanged tag. This custom-node package ZIP is distinct from the official workflow ZIPs.
<img width="1026" height="756" alt="{56C36D94-39EC-4484-9495-5648A1F486F3}" src="https://github.com/user-attachments/assets/ad9bd73f-1ac2-4925-80dd-55a5d3231431" />
## 3.9.6 — regeneration selections and automatic variations
Explicit **Regenerate From** and **Variation Nonce** survive V3.8/V3.9 workflow save/reload. With progress saving enabled, **Full Run + an explicit Chunk + Nonce 0** reaches Run Storage on every Queue, allowing another automatic variation after a completed Take. Earlier accepted chunks and previous Takes are kept. Restart ComfyUI and refresh the browser. For an older unmarked Chunk 1 / nonce 0, select Chunk 1 again and save once. See [operation, compatibility and verified scope](REGENERATE_INTENT_V396.md).
## 3.9.5 — Issue #31: reusable Attention and Spectrum contracts
Review Each Chunk can resume new Runs using Core Model Attention Backend with PyTorch or Comfy Kitchen INT8 and an explicitly audited Spectrum binding, including upstream Spectrum v0.2.29. Existing unsafe history remains readable and read-only for reuse; start a new Run after updating. Public node IDs, workflows and Sampling v6 are unchanged. See [supported implementations, update steps and validation limits](ISSUE31_SAMPLING_SIGNATURES.md).
## 3.9.4 — Registry packaging repair (2026-10-14)
Include the declared Review restoration guide and exclude repository-only prompt-authoring files from the Registry ZIP. The publishing workflow now builds the actual package and checks its exact file set, CRC and every Registry Manifest hash before uploading. Runtime behavior, twelve public node IDs, workflows and saved Runs/Takes are unchanged. GitHub Releases 3.9.3 and 3.9.2 remain available.
## 3.9.3 — choose temporary storage (2026-10-13)
Choose where Continuum writes its temporary decoded frames and Finalize buffer. Open **Settings → H3Continuum → Storage → Manage temporary storage**. Enter an absolute folder on the **ComfyUI server**, or choose a server storage candidate, then validate and save. Leave it blank to keep the existing locations. No workflow changes are required; continue using the V3.9 nodes and your 3.9.1 / 3.9.2 workflows.
- The selected folder applies to **both Finalize and Decode Cache Helper**. It contains a dedicated `h3-continuum-scratch` directory. Saved Runs/Takes and finished videos keep their existing locations.
- The panel shows current and previous locations, temporary file sizes, free space, and files in use or eligible for deletion. **Delete unused temporary files** protects live outputs, Tensor views, active processes and files whose ownership cannot be verified. Deleting a cache can require decoding and writing it again.
- Save or clean only when the running and pending queues are empty. New files use the new location; existing outputs remain valid at their old locations. A missing selected drive does not silently redirect writes to your original SSD. Shared storage controls are read-only in multi-user mode.
- Mode selection stays in the existing nodes: **Finalize: Auto / RAM / Disk-backed**; **Decode Cache Helper: Auto / RAM / Off**. Auto keeps its conservative 4 GiB final-IMAGE threshold and RAM reserve. Manual RAM now checks available memory before allocation; manual Disk-backed checks free space. These checks cannot guarantee against OOM, changing disk usage or a drive disconnect during execution. No new VRAM storage mode is added.
**Moving storage changes where writes occur, not their total volume.** With only one physical drive, moving to another folder or partition on that drive does not divert its writes. A separate SSD is usually the faster alternative; an HDD can be used but may slow decoding, assembly and saving. If sufficient CPU RAM is available, RAM modes can avoid these scratch writes, but memory pressure can cause paging or OOM. Turning Decode Cache Off does not turn off Finalize's Disk-backed buffer. File-size displays are logical sizes, not physical disk allocation or cumulative SSD writes. Local filesystems are supported; network shares are outside this feature's supported scope.
Update the custom node package, restart ComfyUI to load the backend, and hard-refresh the browser. Later changes to this storage setting take effect without a restart. Blank settings retain the ComfyUI temp directory for Finalize and `H3_DECODE_CACHE_TEMP_DIR` (or OS temp) for Decode Cache. A selected folder takes precedence over that cache environment variable.
The storage changes preserve frame dtype, video/audio assembly and generation identity. Validation and remaining platform limits are recorded in [PACKAGE_VALIDATION.txt](../PACKAGE_VALIDATION.txt). The implemented settings screenshot appears at the top of this README. GitHub Release and Registry publication status are listed above; neither follows automatically from a source update.
If the panel reports HTTP 404, inspect the ComfyUI startup log. An old process can keep the chosen address/port occupied, so the updated process fails to start while its new JavaScript is already visible. Stop the obsolete instance or choose another port, then start the updated backend. **Refresh** retrieves storage status; it does not restart ComfyUI. Failed reads now show the HTTP result and keep mutations disabled until status is available.
## 3.9.2 — Issue #30 frontend hotfix (2026-10-13)
This hotfix reduces unnecessary Review and Reference Images UI work:
- Skip upstream Review settings snapshots when no comparison baseline exists.
- Notify connected Samplers once per Reference selection, preserving immediate native values and the original widget callback.
- Skip Reference node resizing when its target dimensions already match its actual dimensions.
Review monitoring with a baseline and fresh Take/Queue checks remain in place. Sampling, saved Runs/Takes, public node IDs and official workflow bytes are unchanged. Keep using the V3.9 nodes.
The frontend repair passed **1,703 CPU/JavaScript regression tests, 2 skipped**, with **2 explicit expected failures** for the independent Reference Plan Inspector bug after repeated Sampler configuration. Offline Core registration/planning and isolated Chromium checks passed. Real ComfyUI browser checks for this package are recorded in [PACKAGE_VALIDATION.txt](../PACKAGE_VALIDATION.txt); mobile freeze resolution is unconfirmed. Reference table rebuilding and Review snapshots with a baseline remain. No new GPU generation is part of this hotfix's validation.
After updating, **hard-refresh the browser (Ctrl+F5)** to load the JavaScript. Restart ComfyUI normally to update the displayed package version. A 3.9.1 Run/Take does not require migration for this UI hotfix. Versioned downloads are listed in [GitHub Releases](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases); Registry publication is separate.
## 3.9.1 — what changed (2026-10-04)
This patch fixes Second Pass conditioning, Take reuse and Review audio:
- **Second Pass:** preserve audio-only/mixed keyframes and inherit the verified conditioning and Reference assignments of the actual Review output, including a single middle chunk.
- **Take reuse:** include sampler closures and MODEL CFG/wrappers in compatibility checks to prevent reuse under different generation settings.
- **Review Driving Audio:** cut the source PCM once at the physical group's natural time, with the same start position for Exact ON/OFF.
- **Audio resampling:** prefer ComfyUI Core's standard API, with the legacy fallback for older Core versions. The Reference Encode Cache limitation for direct VAE weight changes is documented below.
**Updating from 3.9.0:** restart ComfyUI and refresh the browser after updating. Existing v5 Takes and history remain readable and are not deleted, but cannot be reused as prefixes for new v6 generation. Start a new Full Video or choose **Start again from Chunk 1**. Public node IDs, widget order and official workflows remain unchanged; keep using the **V3.9** nodes.
The repaired code passed **1,696 CPU tests / 1 skipped**, targeted Review Second Pass GPU checks and an integrated generation/reuse check. Browser acceptance remains pending. Optional latent-upscale doubled outlines are tracked separately in [Issue #27](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/27); they are not a required gate for these fixes. This is a main-source update, not a new GitHub Release or Registry publication.
## 3.9.1 repair details
Second Pass preserves audio-only keyframes and adapts only the video component of mixed keyframes. Review Driving Audio uses the retained natural frame range of the physical decode group, measured before projection. Exact ON/OFF share that PCM origin. Completely empty Review intervals receive silence for the selected video duration; partially short audio stays short with Exact OFF.
New generation uses Sampling Contract v6 and Graph Contract v4, including sampler code/closures and MODEL CFG/wrapper settings. Saved v5 Takes, raw tensors and provenance remain readable, but cannot be reused in v6 generation. Start a new Full Video or use Review's **Start again from Chunk 1** to create a new revision. Old Takes remain intact. Unobservable external settings still allow fresh generation with reuse disabled; incompatible explicit Take/Retry requests are not replaced with a different Take.
Reference/Driving resampling prefers Core's standard API and uses legacy TorchAudio only when that API is absent. Preserved output PCM keeps its original waveform and sample rate.
**Reference Encode Cache limitation:** direct weight mutation inside the same VAE object is not detected automatically. Call `v3.ref_encode_cache.clear_ref_encode_cache()` afterwards or reload into a new VAE object. Unchanged VAE/input pairs continue to hit the cache. Decode Cache Helper reset affects a separate cache and does not clear Reference Encode Cache.
This repair stays within V3.9: public node IDs, socket/widget order, official workflows, Run Storage v3 and State/Session formats are retained. LoRA Plan is not included. CPU checks and browser/GPU acceptance are separate.
V3.9 also projects captured conditioning onto the actual Review output groups before Second Pass. A partial prefix or a single middle chunk can inherit its own conditioning and Reference assignments when the context validates. Full-sequence completion and saved Takes are unchanged; missing or mismatched context keeps the diagnostic prompt-only fallback.
The repaired runtime passed **1696 CPU tests, 1 skipped**, and three targeted GPU cases at **512×512 / 24 fps**: a two-group partial prefix, Chunk 2 alone, and complete three-group reuse. All groups inherited verified conditioning; saved FLAC matched the corresponding source PCM exactly and existing Takes/raw files were preserved. This validates the targeted repair. Browser acceptance remains pending; optional latent-upscale double outlines are a separate investigation in Issue #27. Historical V3.8 releases and tags remain available unchanged.
![H3 Continuum V3.9: fixed Reference Image slots, per-chunk assignments, and one Sampler input](images/v39/v39-feature-summary.png)
*V3.9 is current on `main`: fixed Image 1–9 slots, per-chunk Reference selection, and one Reference Images connection to the V3.9 Sampler.*
## Getting started video
[H3 Continuum V3.9 — Getting Started: Chunks, Prompts & Reference Images](https://youtu.be/AKJxBXaiG6Q) (12:26)
Learn basic First Image setup, chunk settings, Review/Take retries, Timeline prompting, and Reference Images. Includes English narration and subtitles, seven timestamped chapters, version-labeled UI stills, and actual generated examples.
![H3 Continuum V3.8X2 feature card from its original release](images/v39/v38x2-feature-summary-historical.png)
*This V3.8X2 card is retained for comparison. Its “CURRENT MAIN” label describes the time it was made; V3.9 is current now. The V3.8 Sampler and V3.8X2 workflow remain available, but their Reference wiring is not automatically converted to V3.9.*
## External INT Width/Height fix on main (2026-10-01)
Fixed the visible Width/Height input binding in both V3.8 and V3.9. External INT connections now stay aligned with their displayed controls through workflow save/reload and tab switching. Connected controls are read-only; their displayed numbers are saved Manual reserve values, not a prediction of the upstream node's runtime output. Disconnecting restores local editing.
To set exact dimensions from Core **Int** nodes, choose **Size Source = Manual** and connect the Width and Height INT inputs. **First Image** still uses the image and Resolution setting; connected dimensions do not override it or automatically change Size Source. If no active First Image is available, the existing Manual fallback applies. Dimension validation and backend input order/defaults are unchanged.
Validation: **1569 CPU tests passed, 1 skipped**, including 22/22 new frontend regression cases and 60/60 existing Review cases; Chrome save/reload and five tab round trips for both versions. One V3.9 GPU smoke run with external INT values produced exactly **480×640, 96 frames at 24 fps, with 4-second audio** in **107.3 s** (20 sampling steps); the entire saved video decoded successfully. This is a short single-condition check, not long-form or all-model acceptance. Sampling, Run Storage, the earlier Review/cache repairs, official workflow JSONs/ZIPs and historical Releases are unchanged.
Update Continuum from `main`, then **hard-refresh the browser (Ctrl+F5)** to load the new JavaScript. This fix alone does not require a ComfyUI backend restart; restart normally if the update also brings Python changes or new node definitions. This remains a main-source fix, not a new Release or Registry package.
## V3.9 performance fix on main (2026-10-01)
V3.9 could repeat unnecessary prompt-conditioning preparation when **Prompt Format = Fixed** and **no Reference Images are selected**. This repair reuses the existing bounded conditioning cache when its inputs are unchanged. First Image is supported; List/Timeline prompts, selected Reference Images, Reference Audio, Timeline Video and terminal-pair groups keep their existing paths. This is not the Decode Cache Helper and does not skip Sampling.
In the tested `2 × 5 s`, 480×640 First Image workflow, group preparation fell from about **10.76 s to 0.013 s**. In the post-fix same-session comparison (three measured runs per version, after warm-up), median total time was **273.58 s for V3.8 vs 273.78 s for V3.9**; the ranges overlapped. The earlier roughly 6% gap was no longer observed under these conditions. This does **not** establish a speed fix for every workflow, model or GPU, or a GPU-kernel speedup.
Update Continuum from `main` and **fully restart ComfyUI** to load the Python changes. Validation: **1558 CPU tests passed, 1 skipped; all 8 GPU generation runs succeeded** on ComfyUI 0.38.0 / RTX 5060 Ti 16 GB. Seeds, chunk/frame contracts, Run Storage, V3.8X2/V3.9 official workflows and ZIPs, and historical Releases are unchanged. This is a main-branch fix, not a new Release or Registry package.
## Review restoration fix on main (2026-10-01)
Switching to another workflow and back could hide Review actions even though saved Takes still existed. This frontend repair waits for workflow restoration before reading history and ignores stale results from previous graphs. Real generation-setting changes still disable Review/Take application; restoring the original settings re-enables the actions.
Update the custom node from `main`, then refresh the browser to load the new JavaScript. Validation: 60/60 frontend regression cases, 108 focused CPU tests, five Chrome tab round trips, and a Chrome Base Seed edit/restore test; saved Take metadata stayed unchanged. Sampling, Run Storage contracts, V3.8X2/V3.9 workflow files and historical Releases are unchanged. GPU continuation was not tested for this repair. [Details and verification limits](REVIEW_RESTORATION_REPAIR.md).
## Archived implementation notes (pre-V3.8; not the current guide)
The version-labelled sections below are retained temporarily as source history only. They do not describe the V3.8 standard workflow, public surface, or recommended wiring. For an older workflow, use the migration policy and its matching historical Release/tag.
### V3.7 High-Resolution Refinement Foundation
V3.7 adds **H3 Continuum Sampler V3.7** and completes two foundations for resolution-changing Second Pass workflows without changing V3.6 Production defaults or the existing SIGMAS socket.
### Conditioning survives a resolution change
The Conditioning Adapter rebuilds First/Last images, continuation context, and an owned Still Image Guide at the target latent geometry. Physical-group ownership, Long Terminal Merge, Core PackedLayout/RoPE reconstruction, and first-pass Audio passthrough are preserved. Correctness gates passed at 704×704 and 1152×1152; larger canvases still require substantially more memory and processing time.
### RefineSchedule gives refinement ranges explicit meaning
The internal, versioned `RefineSchedule` contract supports `External`, `Full`, `Tail`, and `Partial`. `External` preserves the incoming SIGMAS tensor exactly, while `Tail` and `Partial` select exact ranges from the same source schedule. The accepted internal Tail 6 gate reproduced the six-evaluation suffix and preserved decoded Audio PCM bit-exact. This contract is an implementation foundation rather than a new public widget, and Production defaults remain unchanged.
### Tail 6 cuts Second Pass time by about one third
The largest practical benefit in V3.7 is a shorter high-resolution Second Pass when the internal schedule uses Tail 6. Instead of evaluating all ten points, Tail 6 runs the exact low-sigma suffix that performs the final refinement work.
| Second Pass range | Sampling time |
|---|---:|
| Tail 10 | **100.27 s** |
| Tail 6 | **64.18 s** |
| Saved | **36.09 s / 36.0%** |
In this gate, Tail 6 finished in about 64% of the Tail 10 time: roughly 1 minute 4 seconds instead of 1 minute 40 seconds. A separate Learned Latent Upscaler gate measured `121.17 s -> 66.32 s` for Tail 10 versus Tail 6, about 45% lower, while API total changed from `170.70 s` to `119.61 s`, about 30% lower. These measurements are configuration-specific and are not universal speed guarantees.
`RefineSchedule` does not make each model evaluation faster. It gives Continuum an exact, versioned way to manage and execute the Tail 6 range; the speedup comes from running six evaluations instead of ten. Tail 6 is not enabled automatically, and V3.7 does not change Production defaults.
### Still Image Guide remains Experimental
V3.7 can map one Still Image Guide from an absolute frame to its owning physical group and rebuild it from the original image during a high-resolution Second Pass. Its positioning, Terminal Merge ownership, Run Storage identity, and non-owner isolation passed correctness testing. Core Add Guide uses hard-anchor semantics, however, so a Guide can cause an abrupt trajectory change at the anchor and can redirect later motion. **Still Image Guide is Experimental and remains on Production HOLD.** It should not be treated as a smooth transition control.
### V3.6.1 Maintenance Hotfix
V3.6.1 fixes focused fail-open/compatibility issues without changing the accepted Balanced 22 Masked AV path. Mixed Timeline/List prompts now report `H3C-P105`; Timeline parsing ignores only standalone `---` lines, keeps existing uncovered-chunk fallback, and never stops generation because of prompt syntax.
With `Continuation Backend = Standard` and Audio Continuity enabled, `Balanced — 22 frames` continues to use Masked AV. `Fast — 5 frames`, `Strong — 39 frames`, and `Auto — conservative` now resolve to the accepted Reference Context transport before Run Storage identity is created. The selected UI values are not rewritten, and the status report states the resolved transport and reason. Standard with Audio Continuity disabled remains Masked Video; Compatibility remains Reference Context.
Run Storage now treats every absent Last Frame spelling as the same empty identity. A normal no-Last-Frame run can therefore extend from two to three chunks as `2 reused, 1 generated`, including narrowly compatible V3.6.1 caches whose former final chunk stored `none`. A real connected Last Frame and the Long Terminal Merge atomic pair remain strict. Turning Run Storage Off also resets the hidden Regenerate From and Variation Nonce values to `Auto` and `0` before queueing.
### V3.6 Masked AV Continuation
![H3 Continuum V3.6 Masked AV Continuation](images/v36-masked-av-continuation.png)
V3.6 adds **H3 Continuum Sampler V3.6**. Its Standard continuation backend places the previous finalized Video and Audio latent prefixes directly inside the next H3 target and protects them with Core noise masks. Unlike the V3.5 Reference Context route, it does not append the same prior frames as a separate Reference block, reducing the packed sequence processed by H3.
| Continuation Backend | Purpose |
|---|---|
| `Standard` | Recommended V3.6 path. Balanced 22 + Audio Continuity uses Masked AV; Fast 5, Strong 39, and Auto safely fall back to Reference Context. When Audio Continuity is off, Masked Video is used. |
| `Compatibility` | Advanced fallback using the accepted V3.5 Reference Context behavior. |
Internal transport identifiers are not exposed in the UI. Run Storage identity follows the resolved transport: Balanced 22 Masked AV stays separate from Reference Context, while a non-Balanced Standard fallback safely shares the identical accepted Reference contract. `Regenerate From` treats the Long Terminal Merge logical pair `[2,3]` as one atomic physical sample.
### Accepted behavior
- T2VA, I2VA, Reference Image + Reference Audio, Balanced 22-frame Joint AV, and 3×5-second FL2VA Long Terminal Merge GPU gates passed.
- Protected Video and Audio prefixes finalize bit-exact; generated regions remain owned by the sampler.
- The accepted Reference Image + Reference Audio output passed subjective audio listening with no reported click, dropout, or stereo-positioning defect.
- In the tested FL2VA Terminal Group 2 pairs, Standard removed 2,874 packed rows (`8.01%`) and reduced median Sampling time by `4.06%` versus Compatibility. Performance varies by model, hardware, and workflow.
- `chunk_seconds` now accepts 4.0–30.0 seconds with a 5.0-second default and 0.1-second step. The 5–15 second range remains recommended and validated; longer high-resolution chunks can substantially increase VRAM use and runtime.
In the historical V3.6 package, V3.5.3, V3.5, and V3.4 Node IDs remained registered for saved workflows. V3.5.3 retained its original Reference Context behavior rather than being silently redirected to the V3.6 backend. This is not a V3.8 registration guarantee.
## V3.5.3 Maintenance Hotfix
V3.5.3 is a distribution-integrity maintenance release. It repairs the public Conditioning Bridge workflow's Width/Height links, removes stale orphan link IDs from the published V3.4/V3.5 templates, corrects Hybrid `<Picture N>` warning numbering, and repairs one legacy PNG for strict image decoders.
There are no changes to generation behavior, nodes, sockets, Sampling, Conditioning payloads, Terminal Merge, Assembly, Seam, Run Storage, Prompt/CLIP caching, Video Guide optimization, or V3.4 compatibility. Existing V3.5.x workflows continue to load unchanged.
## V3.5.2 Stabilization & Optimization Update
V3.5.2 adds no new generation mode. It stabilizes the V3.5.x release, removes repeated Prompt/CLIP work when safe, and lowers temporary memory for long Video Guide inputs while preserving existing workflow and generation contracts.
![H3 Continuum V3.5.2 stabilization and optimization results](images/v352-stabilization-optimization.png)
The figure records the accepted optimization measurements. The final packaging gate additionally passed on ComfyUI 0.33.3 with a 172-entry Manifest, including this image.
### What changed
1. **Repeated-run Prompt/CLIP cache** — unchanged T2VA, I2VA, and FL2VA conditioning can reuse CPU-cached Prompt/CLIP results. Reference, scheduled, tokenizer-option, and hook-modified CLIP paths conservatively bypass the cache. Sampling still runs normally.
2. **Video Guide memory optimization** — the complete source remains part of finite validation and SHA-256 identity, but only the required prefix remains stored for VAE/conditioning work.
3. **Stabilization audit** — measured Sampling, Decode, Assembly, saving, optional-feature tax, and retained memory. No speculative Sampling, Driving Audio, Audio hash, Assembly, Seam, Session, or V3.4 optimization was adopted.
### Measured results
| Area | Before | V3.5.2 | Result |
|---|---:|---:|---:|
| T2VA repeated Prompt/CLIP | 5.898398 s | 0.000028 s | Cache HIT, `encode_calls=0` |
| FL2VA repeated Prompt/CLIP | 21.642687 s | 0.007128 s | Cache HIT, `encode_calls=0` |
| Video Guide peak additional RSS | 396.8 MiB | 114.7 MiB | About 71% lower |
| Video Guide retained storage | 225 MiB | 93 MiB | About 59% lower |
Prompt/CLIP figures measure only the conditioning subphase, not total generation time. The Video Guide memory benchmark uses deterministic 300×256×256 RGB input with a 124-frame required prefix. A separate GPU A/B used an actual 15-second / 360-frame Video Guide and produced identical decoded video and audio PCM.
The measured Sage-only production baselines on the tested RTX 5060 Ti 16 GB / 64 GB system were 168.069 seconds for 1×5-second 576×576 T2VA and 379.765 seconds for 3×5-second 640×640 FL2VA Long Terminal Merge. These are configuration-specific baselines, not universal speed guarantees. Sampling remained the dominant cost; Continuum Assemble + Seam stayed below 1%.
> **V3.8.0 is the historical release baseline.** V3.8X2 remains a separate package-3.8.3 workflow, while V3.9 source is available on `main`. Historical implementation modules remain in source because V3.8X2 reuses them internally. The current registration has ten V3.8X2 IDs plus two V3.9 IDs. Use the matching historical Release/tag for workflows requiring other IDs. Still Image Guide remains Experimental.
## V3.5.1 Reference Audio & Compatibility Update
V3.5.1 added two focused features without changing the V3.4 Sampling, Conditioning, Terminal Merge, Assembly, Seam, or Run Storage contracts:
1. **Optional Reference Audio** — native H3 audio conditioning that does not replace the generated final audio.
2. **Conditioning Bridge V3.5** — one complete Core-compatible `MODEL` and `CONDITIONING` object per Continuum physical group for external sampler workflows.
The Reference Audio sockets are permanently defined by Python `INPUT_TYPES`. Dynamic socket changes and the UI-only `Hidden / Show` control were removed to prevent workflow save/reload value shifts. Node IDs, backend keys, Sampling, Conditioning, standard Seed handling, and other widgets are unchanged.
In the historical V3.5.1 package, V3.4 node IDs and backend socket keys remained registered for saved-workflow compatibility. V3.4/V3.5 workflows continued to load there; V3.5.1 clarified the displayed input names without rewriting saved links. V3.8 instead uses the seven-node support boundary above.
### Reference inputs at a glance
The displayed names intentionally describe different jobs. They are not interchangeable pairs:
| Displayed input | Connect from | Purpose | Final audio behavior |
|---|---|---|---|
| `reference_image_1`–`reference_image_3` | Still-image `IMAGE` | Persistent identity, subject, or appearance references | No audio |
| `Video Guide Frames` | Video loader `IMAGE` frame batch | Persistent motion, framing, timing, and appearance guidance across chunks | Does not carry the video's audio |
| `Driving Audio` + `Driving Audio VAE` | Audio loader, or video loader `AUDIO`, plus the matching audio VAE | Audio guidance whose effective source stream is preserved for final output | Replaces generated final audio with the selected source stream |
| `Reference Audio (Optional)` + `Reference Audio VAE (Optional)` | Standalone audio plus the matching audio VAE | Native H3 conditioning only | Does not copy or replace the generated final audio |
For the common video-with-sound case, connect the loader's `IMAGE` output to `Video Guide Frames` and its `AUDIO` output to `Driving Audio`. Do not connect that audio to `Reference Audio (Optional)` unless conditioning-only behavior is specifically intended.
![Sampler V3.5 permanent Reference Audio, Video Guide Frames, Driving Audio, and Reference Image inputs](images/v351-video-guide-frames.png)
`Reference Audio (Optional)` and `Reference Audio VAE (Optional)` are always present exactly as defined by the node's Python input schema. V3.5.1 no longer adds or removes these sockets dynamically and has no frontend-only `Hidden`/`Show` widget. This keeps positional widget values aligned when workflows are saved and reloaded; backend input keys and existing workflow links are unchanged.
![Sampler V3.5.1 with permanent Reference Audio sockets and aligned widgets](images/v351-reference-audio-permanent.png)
### Conditioning Bridge V3.5
`H3 Continuum Conditioning Bridge V3.5` is the Advanced connection point for external sampling. Connect `model`, `clip`, the externally processed `video_latents`, `assembly_plan`, and `refine_context`; connect `video_vae` only when the selected conditioning path requires it. The node returns parallel `group_models` and `conditioning` lists whose length equals the physical-group count. Each list item remains a complete ComfyUI object—conditioning entries are never flattened into the physical-group list.
```text
H3 Continuum Sampler V3.5.video_latents
-> external H3 latent processor (for example LBH)
-> H3 Continuum Conditioning Bridge V3.5
-> Core BasicGuider + external sampler
-> Core Video / Audio VAE Decode
-> H3 Continuum Assemble + Seam V3.5
```
The external workflow owns AV LATENT pairing, noise, SIGMAS, Audio Lock, sampling, and audio passthrough. The Bridge only exposes the prepared `MODEL` and `CONDITIONING` objects plus the updated Assembly Plan.
![V3.5.1 LBH and Conditioning Bridge external sampling flow](images/v351-lbh-conditioning-bridge-flow.svg)
The retired V3.5.1 LBH + Conditioning Bridge example is available only from its matching historical Release/tag; it is not a V3.8 workflow.
The example keeps the standard Core titles `BasicGuider`, `BasicScheduler`, and `SamplerCustomAdvanced`. Public examples do not rename Core nodes, so they remain immediately distinguishable from Continuum and third-party nodes. The workflow also uses external nodes for LBH latent upscaling, AV LATENT concatenation/separation, loading/saving, and optional acceleration; install or replace those nodes according to your ComfyUI environment.
LBH changes latent geometry only. It has no Continuum denoise-strength control: use the external `BasicScheduler` SIGMAS to decide how strongly the resized latent is regenerated. A larger LBH scale such as 1.5x may still be fast, but memory, decode, and sampling costs increase with the target canvas.
## What's new in V3.5
V3.5 has two major additions:
1. **Continuum-aware Second Pass / Hi-Res Fix** — refine externally processed H3 latents or use the integrated pixel/VAE 2x path without changing V3.4 sampling.
2. **Low-memory Assemble + Seam** — write the final video IMAGE directly to RAM or a Windows-safe mapped file while preserving Exact Duration, Seam, Terminal Merge order, and audio behavior.
### 1. Second Pass and Hi-Res Fix
**H3 Continuum Hi-Res Fix V3.5** is the one-node Main path. It performs Video VAE Decode, bounded pixel resize, Video VAE Encode, and one low-denoise Second Pass. The integrated Main path remains **Experimental** because long 2x runs can exceed GPU memory.
![H3 Continuum Hi-Res Fix V3.5 node](images/v35-hires-fix-node.png)
Normal Hi-Res Fix wiring:
```text
H3 Continuum Sampler V3.5
-> H3 Continuum Hi-Res Fix V3.5
-> Core Video / Audio VAE Decode
-> H3 Continuum Assemble + Seam V3.5
```
**H3 Continuum Second Pass V3.5** is the stable Advanced bridge requested in Issue #8. It accepts an externally processed H3 video latent and samples each Continuum physical group once while preserving prompt policy, seed reporting, group order, and the original first-pass audio LATENT object.
![H3 Continuum Second Pass V3.5 node](images/v35-second-pass-node.png)
External latent processor wiring, such as LBH:
```text
H3 Continuum Sampler V3.5.video_latents
-> external H3 latent processor
-> H3 Continuum Second Pass V3.5
-> Core Video / Audio VAE Decode
-> H3 Continuum Assemble + Seam V3.5
```
Connect the original `audio_latents`, `assembly_plan`, and `refine_context` from Sampler V3.5 to the Hi-Res Fix or Second Pass node. Use the same final MODEL and CLIP, plus a separate sampler and workflow-side SIGMAS for the refinement pass. The nodes do not contain an internal denoise value.
The `BasicScheduler` is deliberately omitted from the screenshots. It is still required to create the Second Pass `SIGMAS`. The GPU-tested baseline is:
```text
sampler: res_multistep
scheduler: simple
steps: 10
denoise: 0.35
```
Do not chain Hi-Res Fix and another Second Pass in the standard path: Hi-Res Fix already performs one Second Pass internally. `H3 Continuum Latent Resize V3.5` is an advanced utility, not the recommended Main 2x path; direct high-ratio interpolation of the 24-channel H3 latent produced persistent artifacts in testing.
If Hi-Res Fix is not connected, V3.4 sampling remains on its original path. V3.5 does not add a forced CPU-offload or low-VRAM sampler mode. `Hi-Res Fix enabled=false` is a lazy exact passthrough and does not evaluate its resize, VAE, or Second Pass dependencies.
See [V3.5 Hi-Res Fix and Second Pass workflow guide](V35_HIRES_FIX.md) for exact sockets, bypass behavior, external-upscaler wiring, and limits.
### 2. Low-memory Assemble + Seam
**H3 Continuum Assemble + Seam V3.5** adds manual RAM / Disk-backed selection and a conservative Auto policy. Disk-backed mode writes the final video IMAGE directly to a mapped file instead of rebuilding the entire completed video in anonymous RAM. Audio remains in RAM.
| Backend | Behavior |
|---|---|
| Auto | Uses RAM only when the final IMAGE is at most 4 GiB and physical-memory reserve is sufficient; otherwise selects Disk-backed. Missing memory counters fail safely to Disk-backed. |
| RAM | Existing in-memory output behavior. |
| Disk-backed | Maps the final video IMAGE to a temporary file; video only. Audio remains in RAM. |
Disk-backed mode preserves the backing file while ComfyUI still references the returned Tensor. Downstream nodes can still allocate a full RAM copy, so Continuum's low-memory guarantee applies to its own assembly stage, not arbitrary downstream implementations.
#### Measured memory reduction
The validated stress case assembled a deterministic `1536 x 1536`, 360-frame float32 IMAGE with a final size of `9.49 GiB`. RAM and Disk-backed outputs were video/audio hash-identical.
| Windows process metric | RAM backend | Disk-backed | Reduction |
|---|---:|---:|---:|
| Private memory | 12.41 GiB | 2.91 GiB | **9.50 GiB** |
| USS | 10.29 GiB | 0.80 GiB | **9.49 GiB** |
This is a **system-RAM/private-commitment** reduction during Continuum Assembly, not a claim that sampler GPU VRAM is reduced. Windows RSS includes resident mapped-file pages and was therefore similar between the two backends; private memory and USS are the relevant measurements. Disk-backed I/O can make final assembly slower, but it does not slow the preceding model sampling path.
#### Optional Windows tip: disable pinned memory
For MiniMax H3 on Windows, starting ComfyUI with `--disable-pinned-memory` can substantially reduce host-RAM pressure. In one local RTX 5060 Ti 16 GB / RAM 64 GB check, a `0.4 MP (640 x 640), 2 x 5s` run completed in `5m 23s`, with about `6.1 GB` sampler RSS and `8.6 GB` after assembly, while showing little practical speed difference from normal startup.
```text
--disable-pinned-memory
```
This is an environment-specific tip, not a default requirement or a universal crash fix. Compare the same workflow and seed with only this option changed, and keep it only when memory use improves without a meaningful speed or stability regression. It disables ComfyUI's pinned host-memory path; it does not guarantee that Windows Shared GPU Memory will never be used.
### GPU acceptance at a glance
| Path | Tested condition | Result |
|---|---|---|
| Main Hi-Res Fix | FL2VA 1 x 5s, 576 -> 1152 | PASS |
| Main Hi-Res Fix | Hybrid FL2VA + Reference 1 x 5s, 576 -> 1152 | PASS |
| Advanced Second Pass | Hybrid/Reference 1 x 5s, 576 x 576 | PASS |
| Main Hi-Res Fix | FL2VA 3 x 5s, 576 -> 1152, RTX 5060 Ti 16 GiB | **CUDA OOM at terminal 77T group** |
For the failed 3 x 5s case, First Pass and the 37T Second Pass group completed. The final logical chunks `[2,3]` remained one 77T Long Terminal Merge physical group and exceeded the tested 16 GiB GPU at its first 1152x1152 Second Pass inference. This is a documented resource limit; the Continuum grouping contract did not change.
## V3.4 compatibility baseline
The historical V3.5 package retained V3.4 nodes for saved workflows; it did not replace their Node IDs, public sockets, Sampling, Conditioning, Terminal Merge, Assembly, Seam, or Run Storage behavior. Those historical workflows remain available through the corresponding Release/tag. Current V3.9 exports only its twelve registered IDs, even when older implementation modules remain in source.
![H3 Continuum V3.4 workflow overview](images/v34-workflow-overview.png)
V3.4 focuses on the two reference workflows that are most useful in normal production:
- **Driving Audio**: use an existing audio source as the preserved final audio while it guides the H3 generation across chunks.
- **Video Guide Frames**: use a video loader's IMAGE frame batch as persistent guidance for appearance, motion, framing, and timing without treating it as a frame-by-frame copy.
- **Hybrid FLF + Reference**: use First Frame, Last Frame, and persistent Reference Images together without adding a separate mode or model allowlist.
- **FL2VA Terminal Merge**: keep the final two 5-second FL2VA chunks as one physical 10-second sampling and decode unit for Core-equivalent Last Frame handling.
- **Restartable chunks**: reuse completed chunks with Run Storage and regenerate only the selected part when the generation contract remains compatible.
- **Core compatibility**: remove Continuum-only rejection of unknown upstream/custom nodes and remove the obsolete `strict_compatibility` control.
- **Spectrum interoperability**: use the official H3 Continuum Interop API v1 when Spectrum is installed; Spectrum remains optional.
- **Simpler public interface**: Timeline Video and the earlier timeline-audio paths are hidden from the V3.4 public sampler interface rather than being presented as stable production features.
### Direction change: from timeline generation to preserved references
Earlier development explored Timeline Video and timeline-audio conditioning. Those paths can be useful for experiments, but they are not the default production behavior: chunk boundaries can change visual content, and generated audio can diverge from a supplied song, dialogue, or effects track.
V3.4 therefore prioritizes:
1. **Driving Audio** for cases where the supplied audio should remain the final audio stream.
2. **Video Guide Frames** for cases where a supplied video should guide the generated result without requiring exact frame reproduction.
3. **Chunked generation and Run Storage** for practical retries and long-form work.
This is a usability and reliability decision, not a claim that the experimental timeline paths are impossible. They are hidden from the stable interface while the public workflow stays focused on predictable user-controlled inputs.
## Example workflow
- [V3.9 workflow JSON](../examples/workflows/MiniMax_H3_Continuum_V39.json) and [ZIP](../examples/workflows/MiniMax_H3_Continuum_V39.zip) — current V3.9 graph; the ZIP contains only this JSON
- [V3.8X2 workflow JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2.json) — retained V3.8X2 graph; Spectrum is disabled by default
- [V3.8X2 workflow ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip) — contains both the V3.8X2 and V3.9 JSON files; it is not a custom-node installer
- [V3.8X2 Helper-explicit JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json) and [ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip) — the same graph under an explicit distribution name
The declared Registry payload includes both V3.9 and V3.8X2 graphs; it is **not dependency-free**. Spectrum, rgthree, and KJNodes must be installed separately; ComfyUI-Easy-Use is not required. Historical workflow files remain in GitHub source but are excluded from the Registry payload. For an older saved workflow, use its matching historical Release/tag as described in the [migration policy](V38_RELEASE_AND_MIGRATION.md). Registry packaging validation/publication is separate from this GitHub source update.
## V3.4 feature details
### Driving Audio
Driving Audio is the primary V3.4 audio path.
- Source audio guides each chunk at its absolute sequence position.
- The original effective stream is preserved for final output.
- Generated audio and Audio Seam processing are bypassed while Driving Audio is active.
- Audio is not independently rewritten at every chunk boundary.
- Perfect lip sync is not guaranteed; results still depend on H3, source material, prompts, and sampling.
### Video Guide Frames
`Video Guide Frames` provides a persistent native H3 video guide across all chunks. Despite its ComfyUI `IMAGE` socket color, it expects the frame batch produced by a video loader rather than one ordinary still-image reference.
- It guides motion, framing, timing, and appearance.
- It is not a direct pixel-copy or deterministic identity-transfer system.
- Video and audio are independent. Route a loader's IMAGE output to `Video Guide Frames` and AUDIO output to `Driving Audio`.
- Source decoding and frame-rate conversion remain loader responsibilities. Continuum does not impose an unnecessary forced 24 fps conversion.
| Video Guide Size | Purpose |
|---|---|
| Efficient - 0.4 MP | Default; lower token cost and faster iteration. |
| Balanced - 0.6 MP | More detail at a higher compute cost. |
| Match Output | Largest reference input; potentially much slower and heavier. |
#### Conditional size controls and internal resize
The compact V3.5 UI shows each size control only when its corresponding input is connected:
- Connect any of `reference_image_1` through `reference_image_3` to reveal `Reference Size`.
- Connect `Video Guide Frames` to reveal `Video Guide Size`. Earlier workflows and reports may call this control `Video Reference Size`; V3.5.1 uses the clearer display name `Video Guide Size` while preserving the existing backend key and saved-workflow compatibility.
These controls are therefore not missing from a newly created node; they are hidden until relevant. Reference Images and Video Guide Frames are resized internally before VAE encoding. Video Guide Frames preserve their aspect ratio, align to the H3 canvas, and use Lanczos when reduction is needed. Smaller sources are not enlarged automatically. An external resize node is normally unnecessary, but remains useful for deliberate cropping, forced upscaling of a small source, or a custom resize/upscale algorithm. Video decoding and frame-rate conversion remain the loader's responsibility.
### Hybrid First/Last Frame + Reference Images
You need an hybrid version of MiniMax-H3 to make both I2V and reference images work:
https://huggingface.co/smhfacct/Minimax-H3-fl2va-ref2va-hybrid-models
https://www.reddit.com/r/StableDiffusion/comments/1vm62pj/i_tested_out_the_b2049_hybrid_variant_for_r2v/
V3.4 supports First Frame, Last Frame, and persistent Reference Images in the same Continuum run. The public sockets and controls are unchanged: connect the inputs that the selected H3 model supports.
- Pure FL2VA and pure Ref2VA identity contracts remain unchanged.
- The combined path receives a distinct hybrid identity so Run Storage does not reuse incompatible chunks.
- Prompt/reference mismatches produce guidance warnings rather than a Continuum-only execution stop.
- No model-name allowlist or automatic model replacement is used.
Local checks passed with the B2049 hybrid variant. FL-only models could run, but showed less reliable transitions when FLF and Reference Images were combined; a hybrid-capable model is therefore recommended for this path.
### FL2VA Terminal Merge
For 5-second FL2VA workflows, Continuum keeps the final two logical chunks together as one physical sampling and VAE decode unit.
- `2 x 5s`: one 243-frame physical sample, decoded once, then adjusted to the requested 240-frame output.
- `3 x 5s` and longer: earlier chunks use normal Continuation; the final two chunks use one 260-frame physical sample containing the 22-frame continuation context.
- After the terminal physical decode, the 22-frame context prefix is removed. For `3 x 5s`, this gives `124 + 238 = 362` frames before exact-duration adjustment to 360 frames.
- If the final two prompt sections differ, they are rebased internally to a local `[0-5s]` / `[5-10s]` timeline. Identical prompts retain the accepted shared-prompt path.
- The terminal pair uses one physical seed and one physical decode while Run Storage continues to preserve normal logical chunk entries.
Local GPU comparison confirmed that `2 x 5s` matches Core 10-second FL2VA sampling and terminal decode behavior. A `3 x 5s` Timeline run also passed with three logical chunks, two physical sampling passes, two physical decode groups, and a 15.000-second output. The short still period near the supplied Last Frame was also present in the equivalent Core result and is treated as normal FL2VA convergence rather than a Continuum-specific failure.
### Core-first, permissive execution
V3.4 removes or relaxes Continuum-only rejection rules that blocked otherwise runnable Core H3 experiments.
- Prompt problems prefer warnings and fallback behavior instead of stopping.
- Unknown upstream/custom wrapper classes are not blanket-rejected.
- There is no model allowlist and no automatic model replacement.
- The obsolete strict_compatibility control is removed from the V3.4 interface.
- Current Core PackedLayout behavior is supported without requiring the removed legacy frame_count argument.
Hard stops remain only for states that cannot execute safely, including corrupt latent topology, incompatible assembly data, invalid persisted revisions, or unusable mandatory payloads.
### Cleaner interface
The public nodes focus on normal production controls. Developer diagnostics and detailed reports are available through ComfyUI settings. The current release path uses `H3 Continuum Sampler V3.8`; older sampler and assembler notes below are retained for historical compatibility context.
<img width="389" alt="H3 Continuum Sampler V3.7 with optional Guide input" src="images/v37-sampler-node.png" />
![H3 Continuum Assemble + Seam V3.5](images/v35-assemble-seam-node.png)
### Timeline paths
Experimental Timeline Video and earlier timeline-audio paths remain in historical implementation code but are not exported by V3.8. Open V3.3 workflows with their historical package; new V3.8 workflows should use `Driving Audio` and `Video Guide Frames`.
## Installation
V3.8 is verified against ComfyUI 0.34.2. Historical acceptance records below may mention older ComfyUI versions; they are not the current installation target.
### Updating an existing installation
For a Git checkout, run the update from the custom-node directory:
```powershell
cd ComfyUI/custom_nodes/ComfyUI-H3-Continuum
git pull --ff-only origin main
```
Restart ComfyUI after the update. If the node was installed with ComfyUI Manager, use its **Update** action instead of running `git pull` manually. Do not mix Manager updates and a separate Git checkout for the same installation.
After the backend restart, search for `H3 Continuum Sampler V3.8` with the V3.8X2 workflow, or `H3 Continuum Sampler V3.9` and `H3 Continuum Reference Images V3.9` with the V3.9 workflow. The source registers twelve IDs in total. If they are missing, check the startup console for the `H3 Continuum 3.9.0 loaded` message and any `ComfyUI-H3-Continuum` import error.
Search for H3 Continuum or Continuum in ComfyUI Manager, or install manually:
~~~bash
cd ComfyUI/custom_nodes
git clone https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum.git
~~~
Restart ComfyUI after installation or update.
## V3.5 nodes
### H3 Continuum Sampler V3.5
The V3.5 sampler preserves V3.4 generation behavior and adds `refine_context` as the sixth output. Connect it to V3.5 Hi-Res Fix, Second Pass, or Conditioning Bridge paths that need the original conditioning context. V3.5.1 also provides permanently visible optional Reference Audio sockets.
### H3 Continuum Conditioning Bridge V3.5
Advanced external-sampler bridge. It returns aligned `group_models` and complete `conditioning` objects per physical group together with the updated Assembly Plan. It does not create AV LATENT pairs, noise, SIGMAS, Audio Lock, or external sampling policy.
### H3 Continuum Hi-Res Fix V3.5 (Experimental)
Main convenience node. When enabled, it performs the Video VAE pixel-resize round trip and Second Pass internally. When disabled, video/audio LATENT objects and the Assembly Plan pass through unchanged and the sampling/VAE dependencies remain lazy.
### H3 Continuum Second Pass V3.5
Advanced Issue #8 bridge for an externally processed video latent. It samples once per physical group, reads the physical prompt contract from the Assembly Plan, and returns the original first-pass audio LATENT objects.
### H3 Continuum Latent Resize V3.5
Utility that changes only video-latent H/W. It does not sample. Direct 2x interpolation at large square targets is not the accepted Main Hi-Res Fix method.
### H3 Continuum Assemble + Seam V3.5
V3.4-compatible assembly with Auto, RAM, and Disk-backed video-buffer backends. Exact Duration and seam corrections write directly to the selected target buffer.
## V3.4 nodes
### H3 Continuum Sampler V3.4
Main inputs:
- model, clip, video_vae, sampler, sigmas
- Sequence Prompt
- first_frame, last_frame
- reference_image_1 through reference_image_3
- Video Guide Frames
- driving_audio, audio_vae
Outputs:
- video_latents, audio_latents
- assembly_plan, status
- driving_audio
Visible controls:
- Prompt Format, chunks, chunk_seconds
- width, height, continuity
- base_seed, control after generate
- audio_continuity
- Run Storage
- Reference Size
- Video Guide Size
`chunk_seconds` uses one shared duration for every chunk. The default is 5.0
seconds; 5–15 seconds remains the recommended and validated range. Durations up
to 30.0 seconds are supported, but values above 15 seconds can substantially
increase VRAM use and processing time, especially at high resolution. Per-chunk
variable durations are not supported.
### H3 Continuum Assemble + Seam V3.4
Inputs: images, audio, assembly_plan, and driving_audio.
Controls: Audio Seam and Video Seam.
When Driving Audio is connected, preserved source audio is selected for final output and generated audio seam processing is bypassed.
## Connection order
~~~text
H3 model / CLIP / Video VAE / sampler / sigmas
|
v
H3 Continuum Sampler V3.4
| | |
video_latents | assembly_plan
|
audio_latents
video_latents -> Core VAE Decode ------- images --+
audio_latents -> Core VAE Decode Audio -- audio --+--> H3 Continuum Assemble + Seam V3.4
assembly_plan -------------------------------------+
driving_audio -------------------------------------+
~~~
For a source video with sound:
~~~text
Video loader IMAGE -> Video Guide Frames
Video loader AUDIO -> driving_audio
~~~
The supplied templates use Video Helper Suite for this split. Any compatible IMAGE/AUDIO loader may be substituted.
## Prompt formats
### Timeline
For **Timeline** mode, each time-range header must be written on its own line. Write the scene description on the following line or lines.
Recommended format:
~~~text
[0-5s]
First section: action, camera movement, and scene progression.
[5-10s]
Continue from the exact final state of the previous section.
[10-15s]
Continue naturally without resetting the scene.
~~~
For 10-second chunks:
~~~text
[0-10s]
Describe the first scene.
[10-20s]
Describe the next scene.
[20-30s]
Continue the sequence.
[30-40s]
Describe the final section.
~~~
**Important:** do not place the prompt text on the same line as the time header.
Correct:
~~~text
[0-5s]
Describe the scene.
~~~
Not recommended:
~~~text
[0-5s] Describe the scene.
~~~
The second form is not recognized as a Timeline header. When **Prompt Format = Auto**, it may therefore be interpreted as a Fixed prompt instead, causing the complete text to be reused across chunks.
The time ranges should normally match the configured `chunk_seconds`.
- `chunk_seconds = 5` → `[0-5s]`, `[5-10s]`, `[10-15s]` ...
- `chunk_seconds = 10` → `[0-10s]`, `[10-20s]`, `[20-30s]` ...
- `chunk_seconds = 15` → `[0-15s]`, `[15-30s]`, `[30-45s]` ...
- `chunk_seconds = 30` → `[0-30s]`, `[30-60s]`, `[60-90s]` ...
#### Global preamble and per-chunk prompting
Text placed before the first Timeline header is treated as a **global preamble** and is automatically included in every chunk prompt. This is a good place for information that should apply throughout the sequence, such as subject definitions, overall style, persistent environment details, or other global instructions.
However, the Timeline sections do not need to contain only the changing action or scene description. For complex scenes, multiple characters, or longer chunks such as 10–15 seconds, it can be useful to repeat important details about the subjects, appearance, location, camera, and current scene state inside each section. More self-contained chunk prompts may sometimes give H3 stronger continuity and reduce cut-like transitions or drift.
Continuum carries visual/audio context from the previous chunk, but that does not guarantee that H3 will preserve every semantic detail automatically. Instead of relying only on phrases such as "continue from the previous chunk," describe the intended next action and important continuity details explicitly.
For example:
~~~text
Subject: A woman in a red jacket.
Style: cinematic nighttime photography.
[0-15s]
A woman in a red jacket stands at the nightclub bar, speaking with two people beside her. Medium shot, slow handheld camera movement, crowded dance floor visible behind them.
[15-30s]
The same woman in the same red jacket continues the conversation at the same bar with the same two people. The camera remains close and slowly moves around the group as one person replies.
[30-45s]
The woman turns toward the dance floor while remaining beside the bar. Keep the same clothing, nightclub environment, nearby characters, and continuous camera style.
~~~
The global preamble is optional. If fully self-contained prompts inside every Timeline section give better continuity for a particular model or scene, that is also a valid approach.
### Other prompt formats
**List** prompts use `---` separators.
**Fixed** reuses one prompt for every chunk.
**Auto** detects Timeline, List, or Fixed formatting from the supplied text.
If Timeline syntax cannot be parsed, V3.4 reports a warning and falls back to an applicable prompt mode rather than rejecting an otherwise runnable workflow.
## Run Storage
Run Storage preserves completed chunks and can reuse them after restarting ComfyUI. Regenerate From selects the first chunk to regenerate; earlier compatible chunks remain unchanged.
![Regenerate From](images/v34-regenerate-from.png)
### Keep the first part and regenerate the second part
For a normal T2VA or I2VA run configured as `2 chunks x 5 seconds`, Chunk 1 is the first five seconds and Chunk 2 is the second five seconds.
Before the first generation:
1. Set `Run Storage` to `Save + Auto Resume`.
2. Choose a fixed `Run Name` and keep the Base Seed fixed.
3. Generate the complete ten-second video once. Continuum stores the raw Video/Audio latent for each completed chunk.
To generate a different second part while keeping the first part:
1. Keep the same Run Name, Base Seed, model, LoRA, resolution, sampler, SIGMAS/steps, references, and Continuation settings.
2. Set `Regenerate From` to `Chunk 2`.
3. Leave `Variation Nonce = 0` for automatic new variations after completed regenerations, or choose a different positive value for an explicit variation. A compatible interrupted regeneration at `0` resumes its existing variation; reusing the same positive value does not request a different result.
4. Queue the workflow again.
Continuum loads the stored Chunk 1 instead of Sampling it again, uses its finalized end as the continuation context, and Samples a new Chunk 2. The report should show `1 reused, 1 generated`. The complete output is then decoded and assembled again; the saved first chunk is reused, while boundary Seam processing is evaluated during the new assembly.
If the two parts need different instructions, define them with List or Timeline Prompt syntax before the first generation. Changing the Prompt Plan or other generation-contract inputs after a stored run can create a new revision, in which case the previous Chunk 1 may not be reusable. A run originally generated with `Run Storage = Off` cannot be reused retroactively. `Regenerate From` works at chunk boundaries; it does not edit an arbitrary region inside one chunk.
FL2VA Long Terminal Merge is an important exception: its final two logical chunks are one atomic physical Sampling unit, so those two chunks are regenerated together.
- Use a fixed seed for reproducible resume.
- Changing persistent models, LoRAs, references, prompts, resolution, or sampling settings can create a new revision.
- Unknown upstream node classes no longer cause rejection by themselves; reuse still depends on compatible observable contracts and hashes.
## Spectrum interoperability
Spectrum is optional. Current Spectrum releases officially support H3 Continuum Interop API v1.
- Chunk 1 uses the normal initial path.
- Chunk 2 and later request Actual Prefix 2.
- Successful continuation logs accepted H3 Continuum API v1, actual prefix=2.
- Unsupported Spectrum versions fall back without a private patch.
See [Spectrum v0.2.15 H3 Continuum interoperability](https://github.com/xmarre/ComfyUI-Spectrum-MiniMax-H3#v0215-h3-continuum-interoperability).
Turbo LoRA and Spectrum are not mutually exclusive. Quality and speed remain workflow-dependent.
## Issue and pull-request response
### Issue #3
V3.4 removes blanket rejection of unknown upstream/custom class names. Run Storage evaluates the observable generation contract instead of treating an unfamiliar wrapper as automatically incompatible.
See [Issue #3](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/3).
### Issue #4
V3.4 follows the current Core H3 layout contract and no longer requires the legacy frame_count parameter. The old strict-compatibility toggle is not part of the public interface.
See [Issue #4](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/4).
### Issue #7
V3.4 allows First Frame, Last Frame, and Reference Images to be used together without requiring the separate BigStationW node or introducing a model allowlist. Hybrid presentation ordering and Run Storage identity are handled by Continuum while pure FL2VA and pure Ref2VA behavior remain unchanged.
The combined path was exercised locally with the B2049 hybrid variant. FL-only models were less consistent for this specific combination, so the README recommends a hybrid-capable model rather than rejecting other models in code.
See [Issue #7](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/7).
### Pull request #1 and #2
These older partial pull requests were superseded by the consolidated [Pull request #5](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/pull/5). V3.4 selectively includes the compatibility direction needed for the current Core H3 contract, while the broader upstream synchronization and release-automation scope remains a separate upstream review.
The older pull requests are not represented as merged V3.4 changes.
### Pull request #5
PR #5 is the consolidated upstream-to-fork proposal covering current H3 compatibility, reference-input handling, assembly memory behavior, exact-duration handling, continuation-reference interoperability, and CI/release automation. V3.4 already contains the user-facing stability decisions described above; the PR remains an upstream review item and is not claimed as merged here.
## V3.4 input connection patterns
V3.4 separates the video-guide frame input from the driving-audio input. Choose the connection pattern that matches your source material.
### 1. Audio only
Connect `Load Audio` to `Driving Audio`. Use this when an existing song, dialogue track, or sound effect should remain the final audio. `Video Guide Frames` is not required.
![Driving Audio connection](images/v34-driving-audio-connection.png)
### 2. Video with its own audio
Connect `Load Video (Upload)` `IMAGE` to `Video Guide Frames`. If the uploaded video contains the audio you want to preserve, connect its `AUDIO` output to `Driving Audio` as well.
![Video Guide Frames and embedded Driving Audio connection](images/v34-video-reference-with-audio.png)
### 3. Video and audio from separate sources
Connect `Load Video (Upload)` `IMAGE` to `Video Guide Frames`, then connect a separate `Load Audio` node to `Driving Audio`. Use this when the video guide and the final audio source are different files.
![Separate Video Guide Frames and Driving Audio connection](images/v34-video-reference-separate-audio.png)
Both inputs are optional. Connect `Video Guide Frames` when video guidance is needed, and connect `Driving Audio` when the supplied audio should be preserved in the final output.
### Video Guide Frames frame rate
Use a 24 fps source for `Video Guide Frames`. `Load Video (Upload)` may accept files recorded at 25 fps or another frame rate, but acceptance alone does not guarantee correct temporal alignment with H3. For a non-24 fps source, set `force_rate` to `24` in `Load Video (Upload)`, or convert the file to 24 fps before loading it. If the source is already 24 fps, leave `force_rate` at its default and do not resample it.
## Current validation status
The public-surface suite checks ten V3.8X2 IDs and twelve IDs including V3.9, official workflows and matching ZIP payloads, declared external dependencies, Registry exclusions, preserved V3.8 widget/socket order, and presentation-only `Show Advanced Settings` / `Hide Advanced Settings` behavior. Registry payload hashes are recorded in `REGISTRY_MANIFEST.sha256`; source integrity is recorded in `MANIFEST.sha256`.
**Historical V3.8X2 preparation evidence:** the full CPU suite passed **1,367 passed / 1 skipped / 0 failed**. The prior dedicated GPU Functional Gate passed a `3 × 5 second` Review Each Chunk run: Q1–Q3 generated one physical group at a time, and Q4 reused all three groups with `3 reused / 0 generated`. Q3 and Q4 had identical decoded RGB and PCM SHA-256 values. The separate nine-reference Gate is recorded below. These checks confirm functional execution, prefix reuse, AV reconstruction, and the tested nine-reference path; they are not a full subjective image- or audio-quality rating.
### Nine-reference input gate (0.3 MP reference images)
On the tested NVIDIA GeForce RTX 5060 Ti 16 GB / 64 GB RAM system, a `3 × 5 second` Full Run completed with Reference Images 1–9, `512 × 608` output, Spectrum Off, LoRA Off, Sage Attention, Balanced 22-frame continuity, and test reference images sized to about `0.30 MP` each. Sampling used approximately `15.0–15.5 GiB` of the available `16,311 MiB` VRAM, so this is a high-resource configuration with little headroom on a 16 GB GPU.
Use the nine-reference bundle when the extra appearance or pose evidence is needed and the system has enough VRAM headroom. This single-seed gate confirms execution and basic visual continuity; it does not establish that nine references are universally higher quality than three, nor guarantee operation with higher-resolution references, larger output sizes, different models, or other concurrent GPU workloads.
For nine-reference use on a 16 GB GPU, pre-resize copies of the source images to around 0.30 MP before loading them. Internal conditioning resize remains active, but it should not be treated as a substitute for controlling the memory pressure of nine large source images.
Validation results apply to the tested local source and environment. They do not guarantee that an installed copy is current, that all models/wrappers fit every GPU, or that a different prompt will have the same visual quality.
## Limits
- V3.8 Second Pass / `refine_context` is not supported while a Review sequence is partial. V3.9 supports validated output-group conditioning for partial Review as described above.
- The 16GB GPU Gate passed with small VRAM headroom; memory use depends on the complete environment and workflow.
- Continuation does not guarantee frame-perfect identity or motion.
- Video Guide Frames guides H3; it does not reproduce every source frame.
- Driving Audio preserves selected audio, but visual lip synchronization remains model-dependent.
- FL2VA may settle on the supplied Last Frame before the requested duration ends; equivalent Core runs can show the same behavior.
- Long, high-resolution sequences may exceed system RAM during final decode and assembly even when chunk sampling succeeds.
- Disk-backed assembly lowers anonymous/private memory commitment for the final Continuum IMAGE, but Core Decode and downstream nodes may still allocate large RAM copies.
- Main Hi-Res Fix 3x5 2x is not accepted on the tested RTX 5060 Ti 16 GiB: the 37T group completed at 1152x1152, but the terminal 77T group failed at its first Second Pass inference with CUDA OOM. Reference/Hybrid-specific 1x5 acceptance passed; longer Hybrid/Reference cases remain unverified.
- Direct high-ratio interpolation with Latent Resize can create persistent H3 artifacts; use the Pixel/VAE Main path or an appropriate external H3 latent processor.
- Match Output can be substantially slower than 0.4 MP or 0.6 MP.
- Seam correction may keep the native boundary when a proposed correction is not safer.
- Optional template nodes must be installed, replaced, or bypassed by the user.
## License
See [LICENSE](../LICENSE).
+517
View File
@@ -0,0 +1,517 @@
# H3 Continuum — 日本語の更新履歴・実装資料
[現在の日本語README](../README_JA.md) · [日本語の詳細ガイド](USER_GUIDE_JA.md) · [CHANGELOG](../CHANGELOG.md)
正本`84899c1`(パッケージ3.9.8)の日本語READMEにあった更新説明・旧版画像・検証記録を保持しています。
「現行」「main上」などの表現、古い導入コマンド、版ごとの受入結果は当時の文脈です。
現在の導入案内や、未検証の条件を含む包括的な合格として読み替えないでください。
最新版の入手先と対応範囲は、日本語READMEを参照してください。
<!-- Preserved README sections begin. -->
**最新版パッケージは3.9.8です。** 固定版のGitHub配布は下表を参照してください。Registryの公開状態は別に確認します。
| バージョン | Release | カスタムノード本体ZIP | SHA-256チェックサム |
| --- | --- | --- | --- |
| 3.9.8 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.8) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/ComfyUI-H3-Continuum-v3.9.8.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.8/SHA256SUMS.txt) |
| 3.9.7 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.7) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/ComfyUI-H3-Continuum-v3.9.7.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.7/SHA256SUMS.txt) |
| 3.9.6 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.6) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/ComfyUI-H3-Continuum-v3.9.6.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.6/SHA256SUMS.txt) |
| 3.9.5 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.5) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/ComfyUI-H3-Continuum-v3.9.5.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.5/SHA256SUMS.txt) |
| 3.9.4 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.4) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/ComfyUI-H3-Continuum-v3.9.4.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.4/SHA256SUMS.txt) |
| 3.9.3 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.3) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/ComfyUI-H3-Continuum-v3.9.3.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.3/SHA256SUMS.txt) |
| 3.9.2 | [Release](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/tag/v3.9.2) | [ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/ComfyUI-H3-Continuum-v3.9.2.zip) | [SHA256SUMS.txt](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases/download/v3.9.2/SHA256SUMS.txt) |
## 3.9.8 — KJ Model Preview Overrideの継続修正
監査済みKJ Model Preview Overrideでは、プレビュー画像を保持しても新Runの保存署名を変えず、別QueueでChunk 1を再利用してChunk 2だけ生成できます。対応は固定したKJソース・Core helperと、**VAE/Audio VAE未接続・Tiny VAE `none`**、Core NoPreviews/Auto/Latent2RGBに限定します。KJNodes 1.5.2全般への対応ではありません。更新後はComfyUIを再起動し、ブラウザーを再読込して新Runを作成してください。旧Take・Workflowの削除は不要です。[対応条件・監査ハッシュ・検証範囲](KJ_PREVIEW_OVERRIDE_V398.md)を参照してください。
### 公開版v3.9.8のGPU再検証 — 2026-10-16:PASS
通常ComfyUIに公開タグ **`v3.9.8`/`58fbd855`** を反映して再起動し、公開版464ファイルとの一致、実ログの3.9.8読込、新しいSampling v6契約内のKJアダプターと固定監査プロファイルを確認しました。新Runを2つ使用した計4 Queueの結果です。
| KJ Preview Override併用構成 | 別QueueのChunk 1→2継続 |
| --- | --- |
| Model Attention Backend:PyTorch/Spectrum OFF | **PASS — 1再利用/1生成** |
| Model Attention Backend:Comfy Kitchen INT8+監査済みローカルSpectrum | **PASS — 1再利用/1生成** |
両構成の2回目は画像を保持したKJ MODELをキャッシュ再利用(**8→16枚**)し、Samplerは実行。契約・revision・元Chunk 1 latentのSHA-256を維持し、`resume_safe=true`でした。映像・音声4本は全デコード成功(512×512/24fps、5秒120フレーム→10秒240フレーム)。保護対象170ファイル(保存Workflow 71件・既存Run/Take 46ファイルを含む)は全ハッシュ不変です。
**対応範囲:** 固定したKJ/Core監査プロファイルのみ。ローカルKJソースはupstreamと異なり、KJNodes 1.5.2全般の合格ではありません。Preview OverrideのVAE/Audio VAE未接続・Tiny VAE `none`・Core Auto、RTX 5060 Ti 16 GB、8-step Turbo/Euler/simple、2×5秒。Spectrumは監査済みローカル版のoffline replay、履歴・archiveはsystem RAM、model-aware OFFです。**未確認:** ブラウザーRunボタン/保存再読込、報告者の完全一致graph、別KJ/Spectrum版、SolAttn/preview VAE、カーネル単位の戻り値、主観画質・音質、速度・VRAM比較。保持PNGは32×32で、高解像度プレビュー品質の証明ではありません。既知Inspector問題は残り、旧unsafe/v5 Takeを自動修復・昇格しません。[Issue #31への結果返信](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/31#issuecomment-6209571261)。
## 3.9.7 — 外部入力接続時の再生成キャッシュ修正
ChunksやVariation Nonceを外部ノードから接続した際に、明示したFull Run再生成がCoreのキャッシュで省略される問題を修正しました。前の採用済みChunk、既存Take、公開入力、公式Workflowは維持します。更新後はComfyUIを再起動し、ブラウザーを再読込してください。Workflowの変更や履歴の削除は不要です。 [Release notes](RELEASE_NOTES_V397.md).
**公開版3.9.7でのGPU再検証(2026-10-15):PASS。** 公開タグ(`18df80d`)で再起動後、外部Core INTからChunks=2・Nonce=0を入力し、完全に同一のプロンプトでChunk 2を2回再生成しました。自動Nonceは**1→2**へ進み、両回ともChunk 1を再利用してChunk 2だけを生成。以前のTake・raw latentも保持しました。基準生成を含む動画3本は240フレーム・映像/音声各10秒で全デコード成功。既存Workflow 60件・設定1件・Run/Take 30ファイル・公開版459ファイルのハッシュは不変です。条件はRTX 5060 Ti 16 GB、512×512、24fps、2×5秒、8-step Turbo、PyTorch attention/Spectrum OFF、Core API経由。CUDA/OOM/実行エラーは確認されず、ブラウザー・主観画質・他アクセラレーターの受入範囲は拡張していません。
元の[3.9.2 Source ZIP](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/archive/refs/tags/v3.9.2.zip)は従来のタグから取得できます。本体ZIPと公式Workflow ZIPは別の配布物です。
## 3.9.6 — 再生成指定・Nonce保持と自動Variationの再実行
V3.8/V3.9の**Regenerate From・Variation Nonceを保存・再読込後も保持**します。保存ONで**Full Run+明示Chunk+Nonce 0**を指定すると、QueueごとにRun Storageが処理し、完了後の再実行で新Variationを作成できます。前のChunkと既存Takeは保持します。更新後はComfyUIを再起動し、ブラウザーを再読込してください。旧markerなしChunk 1/Nonce 0は、Chunk 1を選び直して一度保存してください。[操作・互換性・検証範囲](REGENERATE_INTENT_V396.md)を参照してください。
## 3.9.5 — Issue #31:Attention/Spectrumの保存Take継続
CoreのModel Attention Backend(PyTorch/Comfy Kitchen INT8)と監査済みSpectrum Bindingを正確に識別し、新しいRunのReview Each Chunk継続を可能にします。公式Spectrum v0.2.29も専用プロファイルで対応します。以前のunsafe履歴は保持し、再利用可能へ自動変換しません。更新後は新しいRunで生成してください。公開ノード・Workflow・Sampling v6は維持します。[対応実装・更新手順・検証範囲](ISSUE31_SAMPLING_SIGNATURES.md)を参照してください。
## 3.9.4 — Registry配布の梱包修正(2026-10-14)
Manifestに記載したReview復元ガイドを収録し、配布対象外のプロンプト作成資料を除外します。公開前に実際のRegistry ZIPを作り、全収録ファイル・CRC・Manifestの全ハッシュを照合します。生成動作・12公開ノード・公式Workflow・保存済みRun/Takeは変更しません。3.9.3と3.9.2のGitHub Releaseは保持します。
## 3.9.3 — 一時ストレージの保存先設定(2026-10-13)
未圧縮フレームなどの一時データを、指定したストレージへ保存できるようになりました。**設定 → H3Continuum → Storage → 一時ストレージを管理**を開き、**ComfyUIサーバー側**の絶対パスを入力するかストレージ候補を選択して、検証・保存してください。空欄なら従来の保存先を使用します。ワークフローの変更は不要で、V3.9ノードと3.9.1/3.9.2のワークフローを引き続き使用できます。
- **FinalizeとDecode Cache Helperの両方**が対象です。指定先の`h3-continuum-scratch`専用領域へ保存します。Run/Takeや完成動画の保存先は別です。
- 現在・以前の保存先、一時ファイル量、空き容量、使用中/削除可能な容量を表示します。**未使用の一時ファイルを削除**は、参照中の出力・Tensorのview・生存プロセス・所有者不明のファイルを保護します。キャッシュ削除後は再デコード・再書き込みが必要になる場合があります。
- 保存・削除は実行中と待機中のQueueが空のときに行います。新しいファイルから保存先を切り替え、既存出力は元の場所で保持します。指定先が使えない場合、元のSSDへ黙って戻しません。マルチユーザー運用では共有保存先のUIは閲覧のみです。
- 使用方式は既存ノードで選びます。**Finalize:Auto/RAM/Disk-backed、Decode Cache Helper:Auto/RAM/Off**。Autoの4 GiB境界とRAM安全余裕は維持します。手動RAMは確保前に空きメモリを、手動Disk-backedは空き容量を確認します。ただし、実行中のメモリ・容量変動やドライブ切断まで防ぐ保証はありません。VRAMを保存領域にする機能は追加していません。
**保存先変更は書き込み先を変える機能で、総書き込み量を減らすものではありません。** 物理ストレージが1台なら、同じ機器の別フォルダー・別パーティションへ移しても書き込みを逃がせません。別のSSDは速度を保ちやすく、HDDも使用できますがDecode・結合・保存が遅くなる場合があります。十分なCPU RAMがあればRAM方式で一時書き込みを避けられますが、余裕がなければページファイルへの書き込みやOOMが起こり得ます。CacheをOffにするだけではFinalizeのDisk-backedを無効にしません。容量表示は論理ファイルサイズであり、物理占有量やSSDの累計書き込み量ではありません。対応先はローカルファイルシステムで、ネットワーク共有は正式対応外です。
更新時はカスタムノードを更新し、ComfyUIを再起動してブラウザーを強制再読込してください。その後の保存先設定の変更には再起動は不要です。空欄時はFinalizeがComfyUIのtemp、Decode Cacheが既存の`H3_DECODE_CACHE_TEMP_DIR`指定またはOSのtempを使います。フォルダーを指定した場合、その指定がキャッシュ用環境変数より優先します。
映像のdtype、動画・音声の結合計算、生成条件の識別は維持します。[実装仕様](TEMPORARY_STORAGE_JA.md)と[検証記録](../PACKAGE_VALIDATION.txt)も参照してください。設定画面のスクリーンショットは[英語README](../README.md)の冒頭にあります。Releaseの公開状態は上表を参照してください。Registry公開は別に確認します。
画面にHTTP 404が出る場合は、ComfyUI起動ログを確認してください。旧プロセスが同じアドレス・ポートを占有していると、更新後の起動が失敗し、新しい画面だけが表示される場合があります。残っている旧環境を終了するか別のポートを選んで、新しいComfyUIを起動してください。**Refresh(更新)**は保存先の状態を再取得するボタンで、ComfyUIの再起動ではありません。取得失敗時はHTTP結果と対応方法を表示し、状態を取得できるまで保存・削除を無効にします。
## 3.9.2 — Issue #30 Frontend hotfix(2026-10-13)
ReviewとReference Imagesの不要なUI処理を削減します。
- Review比較基準がない場合、上流設定snapshotの作成を省略。
- Reference選択時のSampler通知を1回へ整理。native値の即時更新と元widget callbackを維持。
- Referenceノードの目標寸法と実寸が同じ場合、不要なサイズ変更を省略。
比較基準がある場合の監視と、Take/Queue時の最新設定確認を維持します。Sampling、保存済みRun/Take、公開Node ID、公式Workflowのbytesは変更しません。引き続きV3.9ノードを使用してください。
Frontend修正で**CPU/JavaScript回帰1703件成功・2件スキップ**、独立したInspector再configure問題の**2条件を期待失敗として記録**しました。Coreのオフライン登録・計画確認と隔離Chromium検証も通過しています。このパッケージの実ComfyUIブラウザー確認は[PACKAGE_VALIDATION.txt](../PACKAGE_VALIDATION.txt)に記録します。モバイルでのフリーズ解消は未確認です。Reference表の再構築と、比較基準がある場合のReview走査は残ります。今回のhotfixで新たなGPU生成は行っていません。
更新後は**ブラウザーを強制再読込(Ctrl+F5)**してください。表示するパッケージ番号も更新する場合は、通常の手順でComfyUIを再起動してください。3.9.1のRun/Takeは、このUI修正のために移行する必要はありません。固定版は[GitHub Releases](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases)を参照してください。Registry公開は別作業です。
## 3.9.1で変わった点(2026-10-04)
Second Pass、Takeの再利用、Review音声に関するバグを修正しました。
- **Second Pass**:音声のみ/複合キーフレームを保持し、途中Reviewや中間Chunk単独でも、その出力区間のconditioningとReference割当を照合して引き継ぎます。
- **Takeの再利用**:Samplerのclosure、MODELのCFG/wrapperも互換性判定に含め、異なる生成設定のTakeを誤って再利用しないようにしました。
- **ReviewのDriving Audio**:物理groupの自然時間に合わせて元PCMを1回だけ切り出します。Exact ON/OFFで開始位置を揃えます。
- **音声resample**:ComfyUI Core標準APIを優先し、古いCoreでは従来方式へ戻します。同じVAE内部の重みを直接変更する場合のReference Encode Cacheの制約も下記に明記しました。
**3.9.0から更新する場合**:更新後にComfyUIを再起動し、ブラウザーも更新してください。旧v5 Takeと履歴は読めて、削除されませんが、新しいv6生成の前半として再利用できません。新規Full Video、または **Start again from Chunk 1** で開始してください。公開Node ID・widget順・公式Workflowは維持し、引き続き **V3.9ノード** を使います。
修正済みコードは **CPU 1,696成功・1スキップ**、対象を絞ったReview Second PassのGPU検証、生成/Take再利用の統合GPU検証に合格しています。ブラウザー受入は未完了です。任意のLATENT拡大で出る二重輪郭は別課題の [Issue #27](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/27) として管理し、今回の修正の必須Gateには含めません。今回もmainソースの更新で、GitHub Release/Registry公開ではありません。
![H3 Continuum V3.9:固定の参照画像番号、チャンク別割り当て、Samplerへの1本の接続](images/v39/v39-feature-summary.png)
*現在の`main`はV3.9です。Image 1~9の固定番号、チャンク別のReference選択、V3.9 Samplerへの1本の接続が新しい操作です。*
## 3.9.1の修正詳細
Second Passで音声のみのキーフレームを保持し、複合キーフレームでは映像だけを空間適応します。ReviewのDriving Audioは、区間選択前の物理decode groupの自然時間から元PCMを切り出します。Exact ON/OFFの開始位置は共通です。完全に空のReview音声だけは選択映像の尺に合う無音を補い、部分的に短い音声はExact OFFで引き延ばしません。
新しい生成はSampling Contract v6/Graph Contract v4を使用します。Samplerのコード・closureとMODELのCFG/wrapper等の設定も再利用判定へ含めます。旧v5のTake・raw・生成履歴は読めますが、v6生成へ継ぎ足せません。新規Full Video、またはReviewの「Start again from Chunk 1」で新しいrevisionを作成してください。旧Takeは残ります。署名を観測できない外部設定でも新規生成は可能ですが、自動再利用は無効になります。明示した不適合TakeやRetryを別Takeへ置き換えません。
Reference/DrivingのresampleはCore標準APIを優先し、そのAPIがない旧Coreだけ従来のTorchAudioを使います。最終出力へ渡す元PCMの波形とsample rateは保持します。
**Reference Encode Cacheの制約**:同じVAEオブジェクトの内部重みを直接書き換えた場合、その変更を自動検知しません。変更後は`v3.ref_encode_cache.clear_ref_encode_cache()`を呼ぶか、新しいVAEオブジェクトへ再ロードしてください。通常の同一VAE・同一入力ではcache HITを維持します。Decode Cache Helperのresetは別のcacheを対象とし、Reference Encode Cacheを消しません。
この修正はV3.9内の変更です。公開Node ID・socket・widget順、公式Workflow、Run Storage v3、State/Session形式を維持し、LoRA Planは追加しません。CPU検証と実ブラウザー・GPU受入は別です。
V3.9では、Second Passへ渡す条件を実際のReview出力区間へ対応させます。途中までの出力や中間Chunk単独でも、条件の照合に成功すれば、その区間自身のconditioningとReference割当を継承します。全体の完了状態や保存済みTakeは変更しません。条件が不足・不一致の場合は、従来の診断付きprompt-only fallbackを維持します。
修正済みランタイムは**CPU 1696成功・1スキップ**、**512×512・24fpsのGPU 3ケース**(途中2区間、Chunk 2単独、完了済み3区間の再利用)で合格しました。全区間の条件継承を確認し、保存FLACと元PCMの対応区間は完全一致、既存Takeとrawファイルも保持しています。合格対象は修正対象の動作です。ブラウザー受入は未完了で、任意のLATENT拡大時の二重輪郭はIssue #27で別途調査します。過去のV3.8 Release/tagはそのまま利用できます。
## V3.9の基本操作を動画で見る
[H3 Continuum V3.9 — Getting Started: Chunks, Prompts & Reference Images](https://youtu.be/AKJxBXaiG6Q)(12分26秒)
First Imageの基本設定、チャンクの設定、Review/Takeでの再試行、Timeline形式のプロンプト、Reference Imagesの使い方を紹介します。英語のナレーションと字幕、7つのチャプターがあり、バージョンを明記したUIの静止画と実際の生成例で説明します。
![公開当時のH3 Continuum V3.8X2機能紹介画像](images/v39/v38x2-feature-summary-historical.png)
*V3.8X2の画像も比較用に残しています。画像内の「CURRENT MAIN」は作成当時の表記で、現在の`main`はV3.9です。V3.8 SamplerとV3.8X2 Workflowは利用できますが、Reference配線はV3.9へ自動変換されません。*
## main更新:外部INTによるWidth/Height入力の修正(2026-10-01)
V3.8/V3.9のWidth/Height入力と表示コントロールの対応を修正しました。外部INTの接続位置をWorkflow保存・再読込やタブ切替後も維持します。接続中の欄は編集不可になり、表示する数値は保存済みのManual予備値です。上流ノードが実行時に出す値を推測して表示するものではありません。接続を外すと手動編集に戻ります。
Coreの **Int** ノードから正確な幅・高さを指定する場合は、**Size Source = Manual**にしてWidth/HeightのINT入力へ接続してください。**First Image**は引き続き画像とResolution設定からサイズを決めます。INTを接続しても優先順位やSize Sourceは自動で変わりません。有効なFirst Imageがない場合は従来のManualフォールバックを使います。寸法検証、Backend入力順序・初期値は変更していません。
確認済み:**全CPU試験1569 PASS/1 skipped**(新Frontend試験22/22・既存Review試験60/60を含む)、両バージョンのChrome保存・再読込とタブ5往復。外部INTを使ったV3.9のGPUテスト1回では、**480×640・24fps・96フレーム・映像/音声とも4秒**を正確に出力しました。20ステップ、全体**107.3秒**で、保存動画の全フレームを正常にデコードできました。短い単一条件の確認であり、長時間継続や全モデルでの受入確認ではありません。Sampling、Run Storage、先行するReview/キャッシュ修正、公式Workflow JSON/ZIP、過去Releaseは変更していません。
Continuumを`main`から更新し、**ブラウザーをCtrl+F5で強制更新**してください。このJavaScript修正だけならComfyUI本体の再起動は不要です。同時にPython修正や新ノード定義も更新された場合は通常どおり再起動してください。今回もmainのソース修正で、新Release/Registry公開ではありません。
## main更新:V3.9の条件付け準備の速度修正(2026-10-01)
**Prompt Format = Fixedかつ有効なReference Imagesがない場合**に、V3.9で余分なプロンプト条件付けの準備が繰り返される経路を修正しました。入力が同じ場合に既存の上限付きconditioningキャッシュを再利用します。First Imageは利用できます。List/Timeline、有効なReference Images、Reference Audio、Timeline Video、終端ペアのgroupは従来の経路を維持します。Decode Cache Helperとは別の修正で、Samplingは省略しません。
検証した`2 × 5秒`・480×640・First Imageの条件では、group準備が約**10.76秒から0.013秒**へ短縮しました。修正後の同セッション比較(warm-up後、各版3回)では、全体時間の中央値は**V3.8が273.58秒、V3.9が273.78秒**で測定範囲も重なり、以前の約6%差は見られなくなりました。全Workflow/モデル/GPUでの解決や、GPU演算そのものの高速化を確認したという意味ではありません。
Continuumを`main`から更新し、Python修正を読み込むため**ComfyUIを完全に再起動**してください。確認済み:**CPU 1558 PASS/1 skipped、GPU生成8回すべて成功**(ComfyUI 0.38.0/RTX 5060 Ti 16 GB)。Seed、チャンク/フレーム契約、Run Storage、V3.8X2/V3.9公式Workflow・ZIP、過去Releaseは変更していません。今回もmainのソース修正で、新Release/Registry公開ではありません。
## main更新:Review表示の復元修正(2026-10-01)
別のWorkflowタブへ切り替えて戻ると、保存済みTakeが残っていてもReview操作が消える問題を修正しました。Workflowの復元完了後に履歴を取得し、古い画面の非同期結果は反映しません。実際に生成設定を変えた場合のReview/Take適用ガードは維持し、元の設定に戻すと操作が復帰します。
カスタムノードを`main`から更新後、ブラウザーを更新して新しいJavaScriptを読み込んでください。確認済み:Frontend回帰60/60、関連CPU試験108件、Chromeのタブ5往復とBase Seed変更/復元。Take管理データは不変です。Sampling、Run Storage契約、V3.8X2/V3.9のWorkflowファイルと過去Releaseは変更していません。この修正のGPU継続生成は未検証です。[詳細と検証範囲](REVIEW_RESTORATION_REPAIR.md)。
## アーカイブされた実装・受入記録(V3.8現行ガイドではありません)
以下は一時的にsource historyとして残る旧リリース記録です。V3.8の標準経路・公開範囲・推奨配線を示すものではありません。旧WorkflowにはMigration Policyと対応するhistorical Release/tagを使用してください。
### V3.7 高解像度Refinement基盤
V3.7では`H3 Continuum Sampler V3.7`を追加し、解像度を変更するSecond Passに必要な2つの基盤を完成させました。V3.6のProduction初期値と既存SIGMAS socketは変更していません。
### 解像度変更後もConditioningを再構築
Conditioning Adapterは、First/Last Image、Continuation context、Still Image Guideの元画像をtarget latent geometryに合わせて再構築します。Physical group ownership、Long Terminal Merge、Core側のPackedLayout/RoPE再構築、First Pass Audioのpassthroughは維持します。704×704と1152×1152のCorrectness GateはPASSしていますが、大きなcanvasほどmemory使用量と処理時間は増えます。
### RefineScheduleでRefinement範囲を明示
内部のversioned contractである`RefineSchedule`は、`External`、`Full`、`Tail`、`Partial`に対応します。`External`は入力SIGMAS tensorをそのまま維持し、`Tail`と`Partial`は同じsource scheduleから範囲を正確に切り出します。内部Tail 6 Gateでは、6 evaluationのsuffix再現とdecoded Audio PCMのbit-exact passthroughを確認しました。現時点では実装基盤として内部で扱い、公開widgetは追加していません。Production初期値も変更していません。
### Still Image GuideはExperimental
V3.7では、Still Image Guide 1枚をabsolute frameからowner physical groupへ割り当て、高解像度Second Passでも元画像から再encodeできます。位置変換、Terminal Merge ownership、Run Storage identity、non-owner groupの非干渉はCorrectness PASSです。一方、Core Add Guideはhard anchorとして働くため、anchor位置で急なtrajectory変化が起き、その後のmotionも別の軌道へ分岐する場合があります。**Still Image GuideはExperimentalで、Production昇格はHOLDです。** 滑らかなtransition制御としては扱わないでください。
### V3.6.1 Maintenance Hotfix
V3.6.1は、受入済みBalanced 22 Masked AV経路を変更せず、限定的なfail-open/互換性問題を修正します。Timeline/List記法が混在したPromptは`H3C-P105`を表示し、Timeline解析ではstandalone `---`行だけを除去します。未指定Chunkの既存fallbackとFixed Promptのfail-open動作は維持し、Prompt構文だけを理由に生成を停止しません。
`Continuation Backend = Standard`かつAudio Continuity ONでは、`Balanced — 22 frames`が引き続きMasked AVを使用します。`Fast — 5 frames`、`Strong — 39 frames`、`Auto — conservative`はRun Storage identity作成前に、受入済みReference Contextへ解決します。UIの選択値は書き換えず、status reportへ解決結果と理由を表示します。Audio Continuity OFFのStandardはMasked Video、Compatibilityは常にReference Contextのままです。
Run Storageは、Last Frame未接続を表す値をすべて同じ空identityとして扱います。通常のLast Frameなし2チャンク生成を3チャンクへ延長すると、旧最終Chunkが`none`だったV3.6.1 cacheを含めて`2 reused, 1 generated`になります。実Last Frame接続時とLong Terminal Mergeのatomic pairは従来どおり厳密です。Run StorageをOffへ変更した場合は、Queue前にhiddenのRegenerate FromとVariation Nonceも`Auto`と`0`へ戻します。
### V3.6 Masked AV Continuation
V3.6では`H3 Continuum Sampler V3.6`を追加しました。標準backendは、前Chunkで確定したVideo/Audio latent prefixを次のH3 Target内へ直接配置し、CoreのNoise Maskで保持します。V3.5のReference Context方式のように、同じ過去フレームを別Reference blockとして追加しないため、H3が処理するpacked sequenceを短縮できます。
| Continuation Backend | 用途 |
|---|---|
| `Standard` | 推奨V3.6経路。Balanced 22+Audio ContinuityはMasked AV、Fast 5/Strong 39/Autoは安全にReference Contextへfallbackします。Audio Continuity OFF時はMasked Videoです。 |
| `Compatibility` | 受入済みV3.5 Reference Context動作へ戻すAdvanced fallbackです。 |
内部transport識別子はUIへ表示しません。Run Storage identityは実行前に解決したtransportへ従います。Balanced 22 Masked AVはReference Contextと分離し、非Balanced Standard fallbackは同一動作のReference contractを安全に共有します。`Regenerate From`でもLong Terminal Mergeのlogical `[2,3]`を1つのphysical sampleとして再利用・再生成します。
### 受入結果
- T2VA、I2VA、Reference Image+Reference Audio、Balanced 22-frame Joint AV、3×5秒FL2VA Long Terminal MergeのGPU GateをPASS。
- 保護Video/Audio prefixは最終的にbit-exact。生成領域はSamplerの出力を維持します。
- Reference Image+Reference Audio代表出力は実聴PASSで、click、dropout、定位異常の報告なし。
- 実測したFL2VA Terminal Group 2では、StandardがCompatibilityよりpacked rowsを2,874行(`8.01%`)削減し、Sampling中央値を`4.06%`短縮。速度はmodel、hardware、workflowで変わります。
- `chunk_seconds`は4.0~30.0秒、初期値5.0秒、step 0.1です。5~15秒を推奨・検証済み範囲として維持し、長時間・高解像度ChunkはVRAMと処理時間が大きく増える場合があります。
V3.5.3、V3.5、V3.4のNode IDと保存済みWorkflowは維持します。V3.5.3は従来Reference Contextのままで、V3.6標準backendへ暗黙に切り替えません。
## V3.5.3 Maintenance Hotfix
V3.5.3は配布整合性だけを修正するメンテナンス版です。公開Conditioning Bridge WorkflowのWidth/Height接続、V3.4/V3.5テンプレート内の孤立link ID、Hybridの`<Picture N>`警告番号、厳密な画像decoderで読めなかった旧PNG 1件を修正しました。
生成動作、Node、socket、Sampling、Conditioning payload、Terminal Merge、Assembly、Seam、Run Storage、Prompt/CLIP cache、Video Guide最適化、V3.4互換性は変更していません。既存V3.5.x Workflowはそのまま読み込めます。
## V3.5.2 Stabilization & Optimization Update
V3.5.2は新しい生成モードを追加する版ではありません。V3.5.xを安定化し、安全な条件で繰り返しPrompt/CLIP処理を省略し、長いVideo Guide入力の一時メモリを削減します。既存Workflowと生成契約は維持します。
![H3 Continuum V3.5.2の安定化・最適化結果](images/v352-stabilization-optimization.png)
画像は受入済み最適化の実測値です。最終パッケージゲートでは、さらにComfyUI 0.33.3と、この画像を含む172項目のManifestでPASSを確認しています。
### 主な変更
1. **繰り返し実行Prompt/CLIP cache**: 条件が変わらないT2VA、I2VA、FL2VAではCPU上のPrompt/CLIP結果を再利用できます。Reference、schedule、tokenizer option、hook変更CLIPは安全側でbypassします。Sampling自体は通常どおり再実行します。
2. **Video Guideメモリ最適化**: source全体をfinite検査とSHA-256 identityへ含めたまま、VAE/conditioningに必要なprefixだけを保持します。
3. **安定化監査**: Sampling、Decode、Assembly、保存、任意機能の負荷、保持memoryを実測しました。根拠のないSampling、Driving Audio、Audio hash、Assembly、Seam、Session、V3.4最適化は採用していません。
### 実測結果
| 項目 | 変更前 | V3.5.2 | 結果 |
|---|---:|---:|---:|
| T2VA繰り返しPrompt/CLIP | 5.898398秒 | 0.000028秒 | Cache HIT、`encode_calls=0` |
| FL2VA繰り返しPrompt/CLIP | 21.642687秒 | 0.007128秒 | Cache HIT、`encode_calls=0` |
| Video Guide Peak追加RSS | 396.8 MiB | 114.7 MiB | 約71%削減 |
| Video Guide保持Storage | 225 MiB | 93 MiB | 約59%削減 |
Prompt/CLIPの数値はconditioning区間だけで、総生成時間ではありません。Video Guideメモリ比較は300×256×256 RGB入力/必要prefix 124 framesの固定CPU条件です。別のGPU A/Bでは実際の15秒/360-frame Video Guideを使い、decoded videoとaudio PCMの完全一致を確認しています。
RTX 5060 Ti 16 GB/RAM 64 GBの検証環境で測定したSage-only Production baselineは、576×576 T2VA 1×5秒が168.069秒、640×640 FL2VA Long Terminal Merge 3×5秒が379.765秒です。環境・設定固有の測定値であり、すべての環境に対する速度保証ではありません。Samplingが最大コストで、Continuum Assemble + Seamは1%未満でした。
**V3.8.0はhistorical release baselineです。V3.8X2はpackage 3.8.3の別Workflowとして維持し、V3.9はmainの現行ソースです。** V3.8X2が内部利用する旧module/classはsourceへ維持します。現在の登録面はV3.8X2の10 IDとV3.9追加の2 IDです。一部は旧IDを維持しますが、それ以外のIDを必要とする旧保存Workflowは対応するhistorical Release/tagを使用してください。Still Image Guideは引き続きExperimentalです。
## V3.5.1 Reference Audio/互換性更新
V3.5.1ではV3.4のSampling、Conditioning、Terminal Merge、Assembly、Seam、Run Storage契約を変更せず、次の2点を追加しました。
1. **Reference Audio(任意)**: 生成音声を置換しない、H3 nativeの音声conditioningです。Workflowの保存・再読込互換を守るため、2つのsocketは常時表示します。
2. **Conditioning Bridge V3.5**: Continuumのphysical groupごとに、完全なCore互換`MODEL`と`CONDITIONING`を外部Samplerへ公開します。
Reference Audio socketはPython `INPUT_TYPES`で常設し、保存・再読込時の値ずれを防ぐため動的socket変更とUI-onlyの`Hidden`/`Show` controlを削除しました。V3.4のNode IDとbackend socket keyは保存済みWorkflow互換のため維持しており、既存V3.4/V3.5 Workflowの接続は変わりません。
### Reference系入力の違い
表示名は役割の違いを示しています。名前が似ていても同じ組み合わせではありません。
| 表示入力 | 接続元 | 用途 | 最終音声 |
|---|---|---|---|
| `reference_image_1`~`reference_image_3` | 静止画`IMAGE` | 人物、被写体、外観などの継続参照 | 音声なし |
| `Video Guide Frames` | Video loaderの`IMAGE`フレームbatch | 全chunkでmotion、framing、timing、appearanceをガイド | 元動画の音声は含まない |
| `Driving Audio` + `Driving Audio VAE` | Audio loader、またはVideo loaderの`AUDIO`と対応Audio VAE | 音声で生成をガイドし、入力音声を最終出力へ維持 | 生成音声を選択した元音声へ置換 |
| `Reference Audio (Optional)` + `Reference Audio VAE (Optional)` | 単独Audioと対応Audio VAE | H3 native conditioning専用 | 生成音声をcopy・置換しない |
音声付き動画を使う通常の接続は、loaderの`IMAGE`を`Video Guide Frames`へ、`AUDIO`を`Driving Audio`へ接続します。conditioning専用動作が目的でない限り、動画の音声を`Reference Audio (Optional)`へ接続しないでください。
![Sampler V3.5の常設Reference Audio、Video Guide Frames、Driving Audio、Reference Image入力](images/v351-video-guide-frames.png)
`Reference Audio (Optional)`と`Reference Audio VAE (Optional)`は、Pythonのinput schemaどおり常時表示します。V3.5.1ではこの2 socketをFrontendで動的に追加・削除せず、UI-onlyの`Hidden`/`Show` widgetも使用しません。これによりWorkflowを保存・再読込した際のwidget値の位置ずれを防ぎます。backend input keyと既存接続は変更しません。
![常設Reference Audio socketと正常なwidget配置](images/v351-reference-audio-permanent.png)
### Conditioning Bridge V3.5
`H3 Continuum Conditioning Bridge V3.5`は外部sampling用のAdvanced接続点です。`model`、`clip`、外部処理後の`video_latents`、`assembly_plan`、`refine_context`を接続し、選択したconditioning経路で必要な場合だけ`video_vae`を接続します。出力`group_models`と`conditioning`はphysical group数と同数で、各要素は完全なComfyUI objectのままです。CONDITIONING内部entryをphysical-group listへflattenしません。
```text
H3 Continuum Sampler V3.5.video_latents
→ LBH等の外部H3 latent処理
→ H3 Continuum Conditioning Bridge V3.5
→ Core BasicGuider + 外部Sampler
→ Core Video / Audio VAE Decode
→ H3 Continuum Assemble + Seam V3.5
```
AV LATENTの組、Noise、SIGMAS、Audio Lock、sampling、audio passthroughは外部workflow側の責任です。Bridgeは準備済み`MODEL`と`CONDITIONING`、更新済みAssembly Planだけを公開します。
![V3.5.1 LBH+Conditioning Bridge外部sampling接続図](images/v351-lbh-conditioning-bridge-flow.svg)
退役したV3.5.1 LBH+Conditioning Bridgeの接続例は、対応するhistorical Release/tagでのみ参照できます。これはV3.8 Workflowではありません。
接続例ではCore標準名の`BasicGuider`、`BasicScheduler`、`SamplerCustomAdvanced`を変更していません。公開サンプルではCoreノードを独自タイトルへ変更せず、Continuumノードや外部カスタムノードと直感的に区別できる状態を維持します。LBH latent upscaler、AV LATENTの結合/分離、load/save、任意accelerationは外部ノードなので、各ComfyUI環境に合わせて導入または置換してください。
LBHはlatent geometryだけを変更し、Continuumのdenoise強度は持ちません。拡大後latentをどれだけ再生成するかは外部`BasicScheduler`のSIGMASで調整します。1.5x等へ上げても高速な場合がありますが、target canvasに応じてmemory、decode、sampling負荷は増加します。
## V3.5の主な追加機能
V3.5の大きな追加は2点です。
1. **Continuum対応Second Pass / Hi-Res Fix**: physical group構造を維持した再Samplingと、標準Pixel/VAE 2x経路。
2. **低メモリAssemble + Seam**: Exact Duration、Seam、Terminal Merge、Audioを維持したRAM / Disk-backed / Auto出力。
### 1. Second Pass / Hi-Res Fix
`H3 Continuum Hi-Res Fix V3.5`はVideo VAE Decode → pixel resize → Encode → low-denoise Second Passを1ノード化したMain経路です。長尺2xではVRAM上限に達する場合があるためExperimentalです。
![H3 Continuum Hi-Res Fix V3.5 node](images/v35-hires-fix-node.png)
通常のHi-Res Fix接続:
```text
H3 Continuum Sampler V3.5
→ H3 Continuum Hi-Res Fix V3.5
→ Core Video / Audio VAE Decode
→ H3 Continuum Assemble + Seam V3.5
```
`H3 Continuum Second Pass V3.5`は外部加工したH3 video latentをphysical group構造のまま再Samplingする安定版Advanced入口です。Issue #8への本体回答です。
![H3 Continuum Second Pass V3.5 node](images/v35-second-pass-node.png)
外部latent upscaler接続:
```text
H3 Continuum Sampler V3.5.video_latents
→ LBH等の外部H3 latent処理
→ H3 Continuum Second Pass V3.5
→ Core Video / Audio VAE Decode
→ H3 Continuum Assemble + Seam V3.5
```
画像では`BasicScheduler`を省略していますが、Second Pass用`SIGMAS`の接続は必要です。GPU受入済み基準は`res_multistep`、`simple / 10 steps / denoise 0.35`、Lanczos pixel resize、FL2VA 1×5秒・576→1152です。Hybrid FL2VA + Referenceも1×5秒で、統合2x Hi-Res FixとAdvanced Second Pass直結の両方を受入しました。
Hi-Res Fixを接続しなければV3.4 Sampling経路は変わりません。強制CPU offloadや低VRAM samplerは追加していません。`enabled=false`ではResize、VAE、Second Passを評価せず、元のVideo/Audio LATENTとAssembly Planをそのまま返します。
### 2. 低メモリAssemble + Seam
`H3 Continuum Assemble + Seam V3.5`はRAM / Disk-backed / Autoを選択できます。Disk-backedでは完成Video IMAGEをWindows対応mapped fileへ直接writeし、AudioはRAMに維持します。
1536×1536・360 frames・完成IMAGE 9.49 GiBの実測では、RAM版とDisk-backed版の映像/音声hashが完全一致しました。
| Windows process指標 | RAM | Disk-backed | 削減 |
|---|---:|---:|---:|
| Private memory | 12.41 GiB | 2.91 GiB | **9.50 GiB** |
| USS | 10.29 GiB | 0.80 GiB | **9.49 GiB** |
これはContinuum Assembly時の**システムRAM/private commitment削減**であり、SamplerのGPU VRAM削減値ではありません。Disk I/Oにより最終Assemblyが遅くなる可能性はありますが、前段のモデルSampling速度は変えません。
### GPU受入範囲
| 経路 | 条件 | 結果 |
|---|---|---|
| Main Hi-Res Fix | FL2VA 1×5秒、576→1152 | PASS |
| Main Hi-Res Fix | Hybrid FL2VA + Reference 1×5秒、576→1152 | PASS |
| Advanced Second Pass | Hybrid/Reference 1×5秒、576×576 | PASS |
| Main Hi-Res Fix | FL2VA 3×5秒、576→1152、RTX 5060 Ti 16 GiB | **Terminal 77TでCUDA OOM** |
3×5秒2xではFirst Passと37T groupのSecond Passは完了し、最後のlogical `[2,3]`を維持した77T physical groupが1152×1152の最初の推論でVRAM上限に達しました。group構造の破損ではなく、16 GiB実機でのリソース上限です。
詳しいsocket接続・OFF時のpassthrough・既知の制限は[英語版V3.5接続ガイド](V35_HIRES_FIX.md)を参照してください。
## 公開テンプレートワークフロー
- [V3.9 Workflow JSON](../examples/workflows/MiniMax_H3_Continuum_V39.json)/[単独ZIP](../examples/workflows/MiniMax_H3_Continuum_V39.zip) — 現行V3.9 graph。ZIPにはこのJSONだけを収録
- [V3.8X2 Workflow JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2.json) — 併存する旧系公式graph。Spectrumは初期状態でOFF
- [V3.8X2 Workflow ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip) — V3.8X2とV3.9の両JSONを格納。カスタムノードのインストーラーではありません
- [V3.8X2 Helper明示名JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json)/[ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip) — 同じgraphを説明的な配布名で収録
Registry配布予定の対象はV3.9とV3.8X2のJSON/ZIPです。依存なしのテンプレートではなく、Spectrum・rgthree・KJNodesが必要です。ComfyUI-Easy-Useは不要です。Turboへ切り替えてもgraph内の外部ノードは残ります。SpectrumをOFFにし、用途に合うLightX2V Turbo LoRAを1本だけONにしてsampler/Stepsを合わせます。保存状態はSpectrum OFF、Turbo LoRA OFF、res_multistep/simple/20 Stepsです。
配布元のprompt、画像・音声名、ノード表示名、設定は変更していません。手元にあるファイルを選び、不要な入力をOFFにし、promptを入力してください。モデルやメディアは同梱しません。`Save 3x5s Video`などの保存済み表示名は生成時間を決めません。実際の長さはSamplerのChunksとSeconds per Chunkで決まります。
旧Workflowは開発履歴としてsource repositoryへ保持しますが、V3.8 Registry packageからは除外します。旧保存Workflowは[Migration Policy](V38_RELEASE_AND_MIGRATION.md)に従い、対応するhistorical Release/tagで開いてください。Registryの正式pack検証・公開はGitHub sourceの更新とは別工程です。
## V3.4互換
V3.4の実装moduleは内部継承とhistorical testのためsourceへ保持しています。現行の登録面は上記12 IDに限定します。それ以外のIDを必要とするV3.4保存Workflowは、対応するhistorical packageを使用してください。
## V3.4.0 Stable
V3.4.0は、最大3枚のReference Image、Run Storage、Spectrum Interop、Decode後のAudio / Video Seam補正を維持し、`Driving Audio`と`Video Guide Frames`を正式な入力として追加します。
`Driving Audio`は入力音声を絶対時間で各Chunkのガイドに使い、最終出力には元音声を維持します。`Video Guide Frames`はVideo loaderが出力したIMAGE frame batchを全Chunkで継続使用し、`Efficient - 0.4 MP`、`Balanced - 0.6 MP`、`Match Output`から処理解像度を選択できます。青いIMAGE socketですが、通常の1枚の静止画参照ではなく動画フレーム列を接続します。
### Size項目の条件付き表示と内部Resize
V3.5のCompact UIでは、対応する入力を接続した時だけSize項目を表示します。
- `reference_image_1`~`reference_image_3`のいずれかを接続すると、`Reference Size`を表示します。
- `Video Guide Frames`を接続すると、`Video Guide Size`を表示します。以前のWorkflowやIssueで`Video Reference Size`と記載されている項目と同じもので、V3.5.1では表示名だけを`Video Guide Size`へ明確化し、既存のbackend keyと保存済みWorkflowの互換性を維持しています。
したがって、新規ノードで見えないSize項目は欠落ではなく、未接続のため非表示になっています。Reference ImageとVideo Guide FramesはVAE Encode前に内部Resizeされます。Video Guide Framesはアスペクト比を維持してH3 canvasへ整列し、縮小が必要な場合はLanczosを使用します。小さい入力を自動的に拡大はしません。通常は外部Resizeノードを必要としませんが、意図的なcrop、小さい素材の強制upscale、独自のresize/upscale方式を使う場合には外部処理を利用できます。動画のdecodeとframe-rate変換は引き続きloader側の責任です。
Reference入力のバイパスされたソケットは無視され、有効な画像だけがPicture 1から連続採番されます。空欄や不完全なプロンプトも停止せず、警告とFixed fallbackで生成を続行します。未知のモデル、ノード、プロンプト形式を独自判断で拒否する制限も撤廃しました。
## Run Storage:第2部分だけを再生成
通常のT2VA/I2VAを`2 chunks x 5 seconds`にした場合、Chunk 1が最初の5秒、Chunk 2が次の5秒です。
初回生成前に次を設定します。
1. `Run Storage`を`Save + Auto Resume`にします。
2. 固定の`Run Name`を選び、Base Seedも固定します。
3. まず10秒全体を一度生成します。Continuumは完了した各Chunkのraw Video/Audio latentを保存します。
最初の5秒を保持して、第2部分だけを作り直す手順は次のとおりです。
1. Run Name、Base Seed、model、LoRA、解像度、sampler、SIGMAS/steps、Reference、Continuity設定を初回と同じに保ちます。
2. `Regenerate From`を`Chunk 2`にします。
3. `Variation Nonce = 0`なら、完了済みの再生成に対して新しいバリエーションを自動選択します。互換性のある未完了runを再開する場合は、既存のバリエーションを引き継ぎます。`1以上`は指定値を固定するため、同じ入力で別の結果を求める場合は別の正の値へ変更します。Base Seedは固定のままにします。
4. Workflowをもう一度Queueします。
Review画面の`Try this chunk again`は、この手動指定とは別に、内部のバリエーションを自動で進めます。
Continuumは保存済みChunk 1を再Samplingせずに読み込み、その確定済み末尾をContinuation contextとして新しいChunk 2をSamplingします。Reportの目安は`1 reused, 1 generated`です。その後、10秒全体を再Decode/Assemblyします。第1ChunkのSampling結果は再利用され、境界のSeam処理は新しいAssembly時に再評価されます。
第1部分と第2部分で異なる指示を使う場合は、初回生成前にListまたはTimeline Promptとして分けておきます。保存後にPrompt Planや生成Contractの入力を変更すると新しいRevisionになり、以前のChunk 1を再利用できない場合があります。初回を`Run Storage = Off`で生成した場合も、後から遡って再利用することはできません。`Regenerate From`はChunk境界単位であり、1つのChunk内部の任意領域を編集する機能ではありません。
FL2VA Long Terminal Mergeは重要な例外です。最後の2つのlogical Chunkを1つのatomicなphysical Samplingとして扱うため、その2つは一緒に再生成されます。
Run Storageの保存契約にはDriving AudioとVideo Guide Framesのidentityを含め、入力変更後に古いChunkを誤再利用しないようにしています。
## Driving Audio / Video Guide Frames / Seam
Driving Audioだけを使う、Video Guide Framesだけを使う、Video Guide Framesと別音声を組み合わせる、という3通りの接続に対応します。Video Guide Framesは24 fps入力を基準とし、異なるfpsの素材はLoad Video側で24 fpsへ設定してください。
H3 Continuum Assemble + SeamはAudio Seam AutoとVideo Seam Autoを既定とし、フレーム削除を行わずに境界の一時的な露出・色変化を補正します。Auto 2は露出ランプ向けの実験的な追加モードです。
## Legacy V2.1.7 Stable Facade UI
V2.1.7では、V2.1.6の動的表示制御を廃止し、標準ComfyUI描画だけを使用する静的Facadeへ移行しました。生成処理は互換用`H3ContinuumSamplerV2`を再利用し、State、Session、Prompt Plan、Spectrum Interop、保存Schemaは変更していません。
通常表示は次を中心にします。
```text
model / clip / video_vae / audio_vae / sampler / sigmas
Sequence Prompt
first_frame
prompt_overrides(任意pack)
advanced(任意pack)
Prompt Format
chunks
chunk_seconds
width / height
continuity
base_seed
control after generate
Seam Correction
Seam Correction
outputs: images / audio / result
```
### 補助ノード
- `H3 Continuum Clip Overrides`: Clip別Promptを1本のpackへまとめます。
- `H3 Continuum Advanced`: resume/session入力と低頻度設定をまとめます。
- `H3 Continuum Result`: resultから`last_state / session / report`を展開します。
- 補助ノードを使わない通常生成では、Sampler本体だけで動作します。
### Legacy互換
当時のV2.1.7では、旧`H3ContinuumSamplerV2`をLegacy/Coreノードとして登録し、既存WorkflowのノードID、入力順、出力順を維持していました。これはV3.8の公開範囲を示す説明ではありません。このIDを使う旧Workflowは対応するhistorical Release/tagで開いてください。
## Continuity
- `Auto — conservative`
- `Balanced — 22 frames`(標準)
- `Fast — 5 frames`
- `Strong — 39 frames (Experimental)`
5/22/39フレームはH3の時間latent周期`1,4,4,4,4`に対して、それぞれ2/7/12 temporal latent stepsです。
## Spectrum
Continuumは有効な前チャンクContextがある場合だけ`h3_continuum` Interop API v1をMODEL optionsへ付与し、対応Spectrumへ`min_actual_prefix_steps=2`を要求します。初回チャンクにはInterop keyを付与しません。未知・未対応Spectrumでは通常Spectrumとしてfail-openします。
## 推奨配線
```text
UNET -> SageAttention -> Sol-Attn -> LoRA -> Spectrum -> H3 Continuum Sampler
```
まず`chunks=2 / chunk_seconds=5 / Balanced 22f / Seam Correction=Off`でNative Continuityを確認し、その後`Seam Correction=Auto`を比較してください。
## インストール
V3.8はComfyUI 0.34.2で検証しています。以下の歴史的な受入記録には旧ComfyUI版も登場しますが、現在のインストール目標ではありません。V3.8の依存を満たす限り、ComfyUIの最新版は必須ではありません。
ZIPを展開して、`ComfyUI-H3-Continuum`フォルダーを`ComfyUI/custom_nodes/`へ置き、ComfyUIを再起動します。
再起動後はV3.8X2 Workflowで`H3 Continuum Sampler V3.8`、V3.9 Workflowで`H3 Continuum Sampler V3.9`と`H3 Continuum Reference Images V3.9`を確認してください。現行の登録面は計12 IDです。
旧`ComfyUI-H3-Continuum-Join`が残っている場合は同時ロードを避けるため削除または退避してください。同梱`install_windows.bat`は旧名・新名の既存フォルダーを日時付きでバックアップします。
## 検査
Public Surface suiteはV3.8X2の10 IDとV3.9を含む計12 ID、公式Workflowと各ZIP内JSONの同一性、外部依存、Registry除外、既存V3.8 widget/socket順、`Show Advanced Settings`/`Hide Advanced Settings`による表示切替を確認します。Registry配布対象のハッシュは`REGISTRY_MANIFEST.sha256`、source側の整合性は`MANIFEST.sha256`で管理します。
**過去のV3.8X2準備証拠:** Full CPU suiteは**1,367 passed / 1 skipped / 0 failed**です。従来の最終GPU Functional Gateでは`3 × 5秒`のReview Each Chunkを実行し、Q1~Q3はphysical groupを1つずつ生成、Q4は`3 reused / 0 generated`で3 groupすべてを再利用しました。Q3とQ4の復号後RGBおよびPCM SHA-256は一致しています。9枚Reference Imageの別Gateは下記に記録しています。これは実行、prefix再利用、AV再構築、テストした9枚経路の確認であり、画像・音声の総合的な主観品質評価ではありません。
### 9枚Reference Image入力Gate(0.3 MP参照画像)
テストしたNVIDIA GeForce RTX 5060 Ti 16 GB/RAM 64 GB環境では、Reference Image 1~9、`512 × 608`出力、`3 × 5秒` Full Run、Spectrum Off、LoRA Off、Sage Attention、Balanced 22-frame continuity、各約`0.30 MP`のテスト参照画像で完走しました。Sampling中のVRAM使用量は、利用可能な`16,311 MiB`のうち約`15.0~15.5 GiB`であり、16 GB GPUでは余裕の小さい高負荷構成です。
9枚bundleは、追加の外観・ポーズ情報が必要で、十分なVRAM余裕を確保できる環境で使用してください。このsingle-seed Gateは実行と基本的な視覚的連続性を確認したものです。9枚が3枚より常に高画質であること、高解像度参照・大きな出力・別model・同時GPU負荷でも動作することを保証するものではありません。
16 GB GPUで9枚を使う場合は、読み込む前に元画像のコピーを約0.30 MPへ縮小してください。内部のconditioning用Resizeは継続しますが、9枚の大きな元画像によるメモリ負荷を管理する代わりにはなりません。
検証結果はテストしたローカルsourceと環境に限られます。導入済み環境が最新版であること、すべてのmodel/wrapperが任意のGPUに収まること、別のpromptでも同じ画質になることを保証するものではありません。
```text
python -m compileall -q .
pytest -q
```
MIT License。
+37
View File
@@ -0,0 +1,37 @@
# Regeneration selections and automatic variations — 3.9.6
## Operation and compatibility
V3.8 and V3.9 retain explicit Regenerate From and Variation Nonce across workflow save/reload and Queue submission. The intent marker is stored in workflow properties; public inputs, widget order and node IDs are unchanged. Choosing Auto clears the marker. The existing one-shot restart remains one-shot.
With progress saving enabled, choose Full Run and an explicit Chunk. Nonce 0 lets Run Storage select an automatic variation: an interrupted compatible variation may continue, and queuing again after completion creates another variation. Fixed positive nonces and ordinary Full Run retain their existing deterministic identity and Core caching.
Restart ComfyUI and refresh the browser after updating. Older unmarked Chunk 1 / nonce 0 values cannot be distinguished from old restart residue; legacy cleanup still restores Auto. To retain a deliberate Chunk 1 / nonce 0, select it again in the updated UI and save once. Old unmarked later-chunk or fixed-positive-nonce selections are retained. Do not delete Runs/Takes to update.
Sampling v6, Graph v4, Run Storage v3 and State/Session formats are unchanged. This patch does not force resume_safe, relax reuse checks, rewrite unsafe/v5 history or migrate workflows. Run Storage decides whether a prefix is compatible. Earlier accepted chunks and previous Takes remain protected. All 25 distributed JSON/ZIP assets retain their bytes.
## Verified scope
Full real-Core CPU gate: 1,908 passed / 2 skipped / 2 existing Inspector expected failures / zero unexpected failures; CUDA initialization attempts 0. Production frontend fixtures cover 84 scenarios including 22 new restoration cases. Real-Core CPU cache checks cover ordinary Full Run, automatic nonce 0 and fixed positive nonces.
Actual normal ComfyUI (Core 0.40.0, frontend 1.55.17): save a separate test workflow, reload the browser, allow real Run history to load, and verify Chunk 2 / nonce 5 and Chunk 1 / nonce 0 remain selected. The saved JSON includes the matching intent marker. This browser test is distinct from actual GPU prompts submitted through the normal Core API.
| New-Run GPU Queue | Result |
| --- | --- |
| Review / Auto / nonce 0 | Generate Chunk 1 |
| Separate Queue, same Review input | Reuse Chunk 1, generate Chunk 2 |
| Full Run / Chunk 2 / nonce 5 | Keep Chunk 1, regenerate Chunk 2 |
| Full Run / Chunk 2 / nonce 0 | Keep Chunk 1, create automatic variation 6 |
| Identical Full Run / Chunk 2 / nonce 0 again | Keep Chunk 1, create automatic variation 7 |
GPU: RTX 5060 Ti, PyTorch attention, Spectrum OFF, 512x512, Euler/simple, 8-step Turbo, two 5-second chunks, no First/Last/Reference inputs. Earlier Chunk 1 SHA-256 stayed identical; every newly generated Chunk 2 differed. The last two prompts, including output prefix, were exactly identical and the Sampler was absent from execution_cached. All manifests recorded Sampling v6 and resume_safe=true. All five video/audio outputs (first: 5 seconds; others: 10 seconds) fully decoded at 24fps.
Existing saved workflows and reachable archived Run/Take files retained their recorded hashes. Older output paths unavailable before this repair were not counted as reverified; no missing-data diagnosis is inferred. Other attention/Spectrum combinations were not newly GPU-tested for this patch. Subjective audiovisual quality, mobile performance, full default-workflow acceptance and the known Inspector reconfigure issue remain separate. Previous Issue #31 coverage: ISSUE31_SAMPLING_SIGNATURES.md.
## 日本語
明示したRegenerate FromとVariation Nonceを保存・再読込後も保持します。保存ON/Full Run/明示Chunk/Nonce 0では、毎QueueでRun Storageが継続か新Variationかを判断します。完了後の同一入力でも再生成し、前のChunkと既存Takeは保持します。固定の正Nonceと通常Full Runのキャッシュは維持します。
ComfyUI再起動とブラウザー再読込が必要です。旧markerなしChunk 1/Nonce 0は、更新後にChunk 1を選び直して一度保存してください。旧unsafe履歴は自動修復しません。既存WorkflowやRun/Takeの削除は不要です。
実画面でChunk 2/5とChunk 1/0の保存再読込保持、実GPUでChunk 1→2継続・Chunk 2再生成・同一Full Run入力2回の自動Nonce 6→7、元Chunk 1保持、動画5本全デコードを確認しました。今回のGPU範囲はPyTorch/Spectrum OFFです。Sampling v6の安全判定、既存配布Workflow、旧Takeを保護しています。
+67
View File
@@ -0,0 +1,67 @@
# 3.9.10 — Spectrum reuse and Reference assignment display
Fix two causes of false **Sampling v6 read-only** in the supported Spectrum
configurations: Windows CRLF source line endings and Python 3.12 implementation
signatures. Also correct the Reference Images assignment table when Chunks comes
from an external INT node.
## Changes
- Audit Spectrum CRLF and LF source line endings identically, including module
closures. All other source bytes and the exact supported file sets stay checked.
- Add separate audited Python 3.12 profiles for upstream Spectrum 0.2.29 at
`806fe498d574a726fecb8818917f00a4b993fd77` and the existing audited local
implementation. Retain the original Python 3.13 profiles byte-for-byte.
- With external Chunks inputs, display **assignment range 1–16** instead of a
stale Sampler widget count. All/Off actions identify their range. This range
does not set generation length or automatically expand saved selections.
- Preserve public node IDs, input/widget order, Sampling v6, Graph v4, existing
Takes and official workflow files. No chunk-LoRA or Project Manager feature is
included.
## Update
Update Continuum, restart ComfyUI and refresh the browser. Existing workflows need
no rewiring. Keep saved Runs/Takes. For a newly supported configuration, start a
**new Run**: older unsafe/v5 Takes remain readable and are not automatically made
reusable. Normal supported Python 3.13 generation identities stay unchanged.
## Verification and limits
- Full local CPU/JS gate: **2,096 passed / 2 skipped / 2 known Inspector expected
failures**; compilation and offline registration of all twelve nodes passed.
- Targeted Spectrum/Attention: **169 passed / 1 skip**. Python 3.12 component:
**112 passed / 1 skip**, with 21 independent older-Core/Attention cases excluded.
- Separate-process storage: **24 cases**, comprising 16 PyTorch/Kitchen cases on
Python 3.13 and eight Spectrum/default-model cases on Python 3.12. All matched
the actual Take read-only JavaScript. Original raw data was retained, and all
six previously valid full Sampling Contracts stayed identical.
- Reference table browser checks passed external INT and KJ Set/Get routing,
selection controls, disconnect/restore, save/reload and workflow tab switching.
- **No new GPU acceptance is claimed for 3.9.10.** These are CPU and UI checks,
not Attention-kernel, H3 inference, image-quality or performance measurements.
Unknown source, changed functions, foreign closures and additional runtime state
remain non-reusable. The repair does not guarantee every Core/Kitchen/KJ/Sage or
Python combination. An independent Attention-observation failure in an older
Core/Kitchen environment can still disable reuse even when Spectrum is recognized.
The reporting user's exact environment has not been reproduced.
See the [compatibility guide](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/blob/v3.9.10/docs/ISSUE31_SAMPLING_SIGNATURES.md) for audit details.
## 日本語
対応済みSpectrum構成で、WindowsのCRLF改行とPython 3.12の実装識別によって
新規Takeまで「Sampling v6 read-only」になる判定を修正しました。外部INTでChunksを
指定したときのReference Images割り当て表も修正し、範囲1–16で設定できるようにしました。
生成チャンク数や既存の選択内容を自動変更する機能ではありません。
更新後はComfyUIを再起動し、ブラウザーを更新してください。Workflowの配線変更や
履歴削除は不要です。新しく対応した構成は新規Runで確認してください。
既存3.13の正常な識別値と、旧unsafe/v5 Takeの閲覧を維持します。
CPU/JS全回帰2,096成功、Python 3.12対象112成功、独立した保存・継続24ケースと
実JS照合、Reference表示のブラウザー操作を確認しました。
**今回の修正としてのGPU受入は未実施です。** 未知の実装や追加状態の拒否は維持し、
すべてのCore/Kitchen/KJ/Sage構成への対応や、報告者本人の環境再現を保証しません。
4.0はメジャーアップデート用として温存し、SemVer互換の3.9.10で公開します。
+21
View File
@@ -0,0 +1,21 @@
# v3.9.5 — Issue #31 Attention and Spectrum Take reuse
Fix finite identity observation of Core Model Attention Backend and slots-based SpectrumH3Binding so compatible new Runs can continue Review Each Chunk. Core PyTorch and Comfy Kitchen INT8 overrides include loaded implementations, closures and both container paths. Spectrum includes every effective config value and an explicit loaded-implementation contract; transient history/counters do not change identity.
Includes a dedicated profile for upstream Spectrum v0.2.29 at 806fe498d574a726fecb8818917f00a4b993fd77, plus the prior audited implementation. Other source/runtime combinations remain non-reusable when their identity is unknown. Restart ComfyUI and start a new Run after updating; previously unsafe/v5 history is preserved and is not promoted or repaired.
Validation: full local CPU gate 1,893 passed / 2 skipped / 2 existing Inspector expected failures / 0 failures; CUDA initialization attempts 0. Targeted signature gate: 152 passed / 1 scope-specific skip. Upstream 0.2.29 PyTorch+Spectrum and Kitchen+Spectrum both persisted Chunk 1 and reused it from a separate Queue to generate Chunk 2, with matching revision/contract, unchanged prefix hashes and fully decoded 512x512, 24fps, 10-second audio/video output. Actual CUDA backend calls were observed in the isolated gate. Prior normal-user-runtime checks also passed for the original audited Spectrum implementation.
All 25 existing JSON/ZIP assets and 259 pre-existing Run/Take files were preserved. Sampling v6, Graph v4, Run Storage v3, twelve public IDs and official workflow bytes remain unchanged. This is targeted functional acceptance; browser interaction, subjective audiovisual continuity, BSA/Flow/Untwist, KJ/Sage patch-node combinations and benchmarks remain separate. See docs/ISSUE31_SAMPLING_SIGNATURES.md for environment-specific support.
GitHub and Registry archives are checked for exact file sets, CRC and complete Manifest hashes. Hosted main and tag tests must pass before publication. Registry scan approval and real Manager installation remain separate from upload/archive verification. Prior tags, Releases and assets are retained.
## 日本語
Model Attention BackendとSpectrumの署名を正確に識別し、新しいRunのReview Each Chunk継続を修正します。公式Spectrum v0.2.29も専用の監査済みプロファイルで対応します。更新・再起動後は新しいRunを作成してください。以前のunsafe/v5履歴は保持し、自動修復・再利用可能への昇格は行いません。
全CPU回帰は1,893成功・2スキップ・既知Inspector問題の期待失敗2件、失敗0・CUDA初期化0。公開Spectrum v0.2.29のPyTorch/Comfy Kitchen両条件で、別QueueからChunk 1を再利用してChunk 2へ継続しました。署名・元Takeのハッシュ保持・完成10秒動画の全デコードと実CUDA呼出しを確認しています。
既存25資産と259件のRun/Takeファイル、公開12ノード、Workflow、Sampling v6を保護しました。未知のSpectrumや異なる実行環境は安全側で再利用を拒否します。実画面・目視画質/音声・BSAやKJの組合せ全般の合格を主張するものではありません。
GitHub本体ZIP・チェックサム、Registry配布ZIPの整合性は公開工程で別々に確認します。Registry側の検査承認と実Manager導入は別判定です。3.9.4/3.9.3/3.9.2などの元タグ・Release・配布物は保持します。
+21
View File
@@ -0,0 +1,21 @@
# v3.9.6 — Regeneration intent and automatic variations
Fix Regenerate From / Variation Nonce restoration in V3.8 and V3.9. Explicit Full Run regeneration with progress saving and nonce 0 now reaches Run Storage on every Queue, so repeating the same input after a completed variation creates another Take while preserving earlier accepted chunks. Ordinary Full Run and fixed positive nonces keep their caching behavior.
Restart ComfyUI and refresh the browser after updating. For an older unmarked Chunk 1 / nonce 0, select Chunk 1 again and save once. Existing Runs/Takes and official workflows do not need deletion or migration. Sampling v6 safety and unsafe/v5 history guards are unchanged.
Validation: 1,908 local CPU passed / 2 skipped / 2 known Inspector expected failures / zero unexpected failures and CUDA initialization attempts. Actual browser save/reload retained Chunk 2 / nonce 5 and Chunk 1 / nonce 0 after history loading. Five new-Run GPU queues passed Chunk 1 creation, separate-Queue Chunk 2 continuation, Chunk 2 regeneration and two identical Full Run / nonce 0 submissions (automatic variations 6 then 7). Earlier Chunk 1 bytes were unchanged; all five video/audio outputs fully decoded.
GPU scope: PyTorch attention, Spectrum OFF, RTX 5060 Ti, 512x512, 8-step Turbo, two 5-second chunks. Browser acceptance and Core-API GPU submissions are distinct checks. Subjective quality, other accelerators and the known Inspector issue are not newly accepted. Existing saved workflows, reachable archived Takes and all 25 JSON/ZIP distribution assets were hash-protected. Previously unavailable old output paths are not claimed as reverified. Details: docs/REGENERATE_INTENT_V396.md.
GitHub ZIP/checksums and Registry archive pass exact file-set, CRC and Manifest checks during publication. Registry upload, service scan approval and actual Manager installation are separate. Earlier Releases/tags/assets are retained.
## 日本語
V3.8/V3.9の再生成指定・Nonceを保存再読込後も保持します。保存ON/Full Run/明示Chunk/Nonce 0の同一入力でもRun Storageへ到達し、完了後は新Variationを生成できます。前のChunkと既存Takeを保持し、通常Full Runと固定Nonceのキャッシュは維持します。
更新後はComfyUIを再起動し、ブラウザーを再読込してください。旧markerなしChunk 1/Nonce 0は選び直して一度保存してください。既存WorkflowやRun/Takeの削除・移行は不要です。Sampling v6の安全判定と旧unsafe履歴の扱いは変更しません。
CPU 1,908成功・2スキップ・既知Inspector期待失敗2、失敗0・CUDA初期化0。実画面でChunk 2/Nonce 5とChunk 1/Nonce 0の保存再読込保持を確認。PyTorch/Spectrum OFFの新RunでGPU 5 Queueが成功し、Chunk 1→2継続・Chunk 2再生成・同一入力の自動Nonce 6→7・元Chunk 1保持・動画5本全デコードを確認しました。画質評価や他アクセラレーターの新規合格は主張しません。
GitHub配布物整合性とRegistryアップロードは公開工程で確認します。Registry検査承認・実Manager導入は別判定です。以前の元タグ・Release・配布物を保持します。
+30
View File
@@ -0,0 +1,30 @@
# v3.9.7 — External-input regeneration cache fix
Fix automatic Full Run variations when Chunks or Variation Nonce come from external nodes. Unresolved linked inputs no longer let Core skip an explicitly requested regeneration. Earlier accepted chunks, existing Takes, public inputs and official workflows are preserved. Update the custom nodes, restart ComfyUI and refresh the browser; no workflow changes or history deletion are required.
CPU: 1,948 passed / 2 skipped / 2 known Inspector expected failures; zero unexpected failures and zero CUDA initialization attempts. Targeted Core-API GPU queues verified external Chunks=2 / Nonce=0, repeated identical automatic variations 1 then 2, and 1 reused / 1 generated chunk per variation. Browser save/reload and Run-button submissions were checked separately. GPU scope: PyTorch attention, Spectrum OFF, 512x512, 8-step Turbo, two 5-second chunks. This is a targeted cache repair, not general image-quality or all-accelerator acceptance.
Known ordinary Full Run, storage OFF and directly specified fixed nonces retain stable caching. An unresolved externally connected nonce may reevaluate the Sampler even when fixed; Run Storage still decides safe prefix reuse. Sampling v6 / Graph v4 and existing unsafe/v5 history safeguards are unchanged.
Existing workflow JSON/ZIP assets remain byte-for-byte unchanged. GitHub asset integrity and Registry upload are verified by the publishing workflow; Registry scan approval and Manager installation remain separate checks.
## 日本語
ChunksやVariation Nonceを外部ノードから接続した際に、明示したFull Run再生成がCoreのキャッシュで省略される問題を修正しました。前の採用済みChunk、既存Take、公開入力、公式Workflowは維持します。更新後はComfyUIを再起動し、ブラウザーを再読込してください。Workflowの変更や履歴の削除は不要です。
CPU 1,948成功、2スキップ、既知の期待失敗2件。外部INT接続で同一入力から自動Variationが進み、Chunk 1を保持してChunk 2だけ再生成することをGPUで確認しました。ブラウザーの保存・再読込とRun操作も別途確認しています。画質・全アクセラレーター構成の新規合格は主張しません。外部接続の固定Nonceはキャッシュ判定時に値が未解決のためSamplerを再評価する場合がありますが、保存済みprefixの再利用は既存の安全判定に従います。
## Post-release verification on the published v3.9.7 tag
**Published v3.9.7 GPU recheck (2026-10-15): PASS.** After restarting with the published tag (`18df80d`), external Core INT nodes supplied Chunks=2 and Nonce=0. Two completely identical Chunk 2 regeneration prompts advanced automatic Nonce **1 → 2**; both reused Chunk 1 and generated only Chunk 2. Earlier Takes and raw latents were retained. All three outputs (baseline plus two variations) fully decoded with 240 frames and 10-second video/audio. Existing 60 workflows, one settings file, 30 Run/Take files and all 459 release files remained hash-identical. Scope: RTX 5060 Ti 16 GB, 512×512, 24 fps, 2×5 seconds, 8-step Turbo, PyTorch attention, Spectrum OFF, Core API submissions. No CUDA/OOM/execution errors were observed; this does not extend browser, subjective-quality or other-accelerator acceptance.
| Queue | Reused chunks | Generated chunks | Automatic Nonce | Elapsed |
| --- | ---: | ---: | ---: | ---: |
| Baseline | 0 | 2 | 0 | 177.8 s |
| Regenerate Chunk 2, first run | 1 | 1 | 1 | 90.4 s |
| Same input, second run | 1 | 1 | 2 | 90.5 s |
### 公開版での追加GPU検証
**公開版3.9.7でのGPU再検証(2026-10-15):PASS。** 公開タグ(`18df80d`)で再起動後、外部Core INTからChunks=2・Nonce=0を入力し、完全に同一のプロンプトでChunk 2を2回再生成しました。自動Nonceは**1→2**へ進み、両回ともChunk 1を再利用してChunk 2だけを生成。以前のTake・raw latentも保持しました。基準生成を含む動画3本は240フレーム・映像/音声各10秒で全デコード成功。既存Workflow 60件・設定1件・Run/Take 30ファイル・公開版459ファイルのハッシュは不変です。条件はRTX 5060 Ti 16 GB、512×512、24fps、2×5秒、8-step Turbo、PyTorch attention/Spectrum OFF、Core API経由。CUDA/OOM/実行エラーは確認されず、ブラウザー・主観画質・他アクセラレーターの受入範囲は拡張していません。
+47
View File
@@ -0,0 +1,47 @@
# v3.9.8 — KJ Model Preview Override continuation
Fix saved-Take continuation after an explicitly audited KJ Preview Override retains preview images. The dedicated adapter separates reviewed transient frames from fixed settings and implementation identity, preserving cross-key/list wrapper execution order. A separate Queue can reuse Chunk 1 and generate only Chunk 2. Sampling v6 safety, existing Takes, public inputs and official workflow bytes are preserved.
**Limited Phase A support:** Preview Override VAE / Audio VAE disconnected, Tiny VAE `none`, Core NoPreviews/Auto/Latent2RGB, and only the exact provider/Core audit profile. KJNodes 1.5.2 alone is not sufficient; the audited local Preview Override source differs from upstream. Unknown or altered providers retain conservative non-reuse. See [the supported source hashes and update guide](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/blob/v3.9.8/docs/KJ_PREVIEW_OVERRIDE_V398.md).
Restart ComfyUI, refresh the browser and start a new Run. Do not delete old workflows or Takes. Earlier safe KJ identities may differ; unsafe/v5 history is not repaired or promoted. Non-KJ v6 signatures are unchanged.
CPU: 1,995 passed / 3 skipped / 2 existing Inspector expected failures; zero unexpected failures or CUDA initialization attempts. Dedicated real-provider tests: 48 passed, no skips. Four normal-runtime Core-API GPU Queues passed: PyTorch and Comfy Kitchen INT8 + audited local Spectrum, each saving Chunk 1 and continuing to Chunk 2 in a separate Queue with held previews. The KJ MODEL was cached on each second Queue, contracts stayed equal, and original Chunk 1 bytes were preserved. All four AV outputs fully decoded.
GPU scope: RTX 5060 Ti 16 GB, 512x512, 24fps, 8-step Turbo, 2x5-second chunks, Core Auto preview, no preview VAE. The initial GPU gate used candidate product bytes; the subsequent published-tag recheck is recorded below. No new browser Run-button, per-kernel return trace, subjective-quality or performance acceptance is claimed. The Inspector issue remains. Source/tag CI and exact GitHub/Registry archive checks run in publication; Registry scan approval and real Manager installation are separate.
## 日本語
監査済みKJ Model Preview Overrideで画像を保持した後、保存Takeを継続できない問題を局所修正しました。一時画像と固定設定を分離し、実装識別とラッパー順序、Sampling v6の安全判定を維持します。
対応は固定した監査プロファイルと、VAE/Audio VAE未接続・Tiny VAE none・Core NoPreviews/Auto/Latent2RGBに限定します。KJNodes 1.5.2全般を保証するものではなく、監査したローカルソースは公開upstreamと異なります。未対応構成は安全側で再利用不可を維持します。
更新後はComfyUIを再起動し、ブラウザーを再読込して新Runを作成してください。既存Take・Workflowは保持し、旧unsafe履歴を自動修復しません。CPU 1,995成功(専用48件を含む)、PyTorchおよびKitchen+ローカルSpectrumの4 Queueで別QueueのChunk 1→2継続、元Chunk 1保持、動画4本の全デコードを確認。画質・速度・全KJ構成の合格は主張しません。Registryの検査承認と実Manager導入は別確認です。
## Post-release verification on the published v3.9.8 tag
### Published v3.9.8 GPU recheck — 2026-10-16: PASS
After restarting the normal ComfyUI installation with the published tag **`v3.9.8` / `58fbd855`**, all 464 installed release files matched the tag. The startup log confirmed 3.9.8, and new saved Sampling v6 contracts contained the KJ adapter and pinned audit-profile digest. Two brand-new Runs passed four Core-API GPU Queues:
| KJ Preview Override configuration | Separate-Queue Chunk 1 → 2 |
| --- | --- |
| Model Attention Backend: PyTorch, Spectrum OFF | **PASS — 1 reused / 1 generated** |
| Model Attention Backend: Comfy Kitchen INT8 + audited local Spectrum | **PASS — 1 reused / 1 generated** |
On both second Queues, the KJ MODEL stayed cached with preview images retained (**8 → 16**), while the Sampler executed. The contract/revision and original Chunk 1 latent hashes stayed unchanged; `resume_safe=true`. All four video/audio outputs fully decoded (5 seconds / 120 frames, then 10 seconds / 240 frames, 512×512 at 24fps). Existing workflows and Takes were preserved: 170 protected files, including 71 saved workflow files and 46 pre-existing Run/Take files, retained their hashes.
**Scope:** only the exact audited KJ/Core profile; the local KJ source differs from upstream, so KJNodes 1.5.2 alone is insufficient. Preview Override VAE / Audio VAE were disconnected, Tiny VAE was `none`, Core preview was Auto. GPU: RTX 5060 Ti 16 GB, 8-step Turbo, Euler/simple, 2×5-second chunks. Spectrum used the audited local offline-replay implementation, system-RAM history/archive and model-aware OFF. **Unverified:** browser Run-button/save-reload behavior, the reporter's exact complete graph, other KJ/Spectrum builds, SolAttn/preview VAE paths, per-kernel return tracing, subjective quality and speed/VRAM benchmarks. Retained PNGs were 32×32, not high-resolution preview-quality evidence. The known Inspector issue remains; old unsafe/v5 Takes are not repaired or promoted. [Issue #31 result](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/31#issuecomment-6209571261).
### 公開版v3.9.8のGPU再検証 — 2026-10-16:PASS
通常ComfyUIに公開タグ **`v3.9.8`/`58fbd855`** を反映して再起動し、公開版464ファイルとの一致、実ログの3.9.8読込、新しいSampling v6契約内のKJアダプターと固定監査プロファイルを確認しました。新Runを2つ使用した計4 Queueの結果です。
| KJ Preview Override併用構成 | 別QueueのChunk 1→2継続 |
| --- | --- |
| Model Attention Backend:PyTorch/Spectrum OFF | **PASS — 1再利用/1生成** |
| Model Attention Backend:Comfy Kitchen INT8+監査済みローカルSpectrum | **PASS — 1再利用/1生成** |
両構成の2回目は画像を保持したKJ MODELをキャッシュ再利用(**8→16枚**)し、Samplerは実行。契約・revision・元Chunk 1 latentのSHA-256を維持し、`resume_safe=true`でした。映像・音声4本は全デコード成功(512×512/24fps、5秒120フレーム→10秒240フレーム)。保護対象170ファイル(保存Workflow 71件・既存Run/Take 46ファイルを含む)は全ハッシュ不変です。
**対応範囲:** 固定したKJ/Core監査プロファイルのみ。ローカルKJソースはupstreamと異なり、KJNodes 1.5.2全般の合格ではありません。Preview OverrideのVAE/Audio VAE未接続・Tiny VAE `none`・Core Auto、RTX 5060 Ti 16 GB、8-step Turbo/Euler/simple、2×5秒。Spectrumは監査済みローカル版のoffline replay、履歴・archiveはsystem RAM、model-aware OFFです。**未確認:** ブラウザーRunボタン/保存再読込、報告者の完全一致graph、別KJ/Spectrum版、SolAttn/preview VAE、カーネル単位の戻り値、主観画質・音質、速度・VRAM比較。保持PNGは32×32で、高解像度プレビュー品質の証明ではありません。既知Inspector問題は残り、旧unsafe/v5 Takeを自動修復・昇格しません。[Issue #31への結果返信](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/31#issuecomment-6209571261)。
+69
View File
@@ -0,0 +1,69 @@
# 3.9.9 — Public KJ Preview Override and Review continuation
Some new Runs were marked **Sampling v6 read-only** when using the upstream KJ
Model Preview Override implementation, even though generation completed. This
release adds a separately audited upstream profile and explains saved reuse
failures in Review and Render History. The original supported local profile and
unaffected generation identities are retained; the safety checks are not relaxed.
## Update
Update Continuum, restart ComfyUI and hard-refresh the browser. Existing workflows
need no rewiring. Keep all saved Runs/Takes. After correcting an unsupported
configuration, start a **new Run**: old unsafe/v5 Takes remain readable and are
not automatically promoted to reusable history.
Supported upstream KJ commit: `17195fcfa8a05f601b766a0f09c9733284f47416`.
On **Model Preview Override**, leave its optional VAE and Audio VAE disconnected,
set Tiny VAE to `none`, and use Core NoPreviews, Auto or Latent2RGB. Keep the
normal generation VAEs connected. This is not blanket support for every KJ/Core
version or unknown wrapper. See [the exact profiles and recovery steps](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/blob/v3.9.9/docs/KJ_PREVIEW_OVERRIDE_V398.md).
## Validation
- CPU: **2,075 passed / 2 skipped / 2 known Inspector expected failures**, zero
unexpected failures. The 90 production-JS cases and eight actual Run Storage
vectors passed; these overlap the pytest total. Offline twelve-node,
PackedLayout and Fixed 3x5s checks passed with zero recorded CPU CUDA
initialization attempts.
- Upstream KJ GPU: two separate Queues, **512x512 / 24fps / 2 x 5 seconds**,
FL2VA INT8 convrot, Qwen3VL NVFP4 AWQ, 8-step Turbo/Euler/simple and explicit
PyTorch attention. Chunk 1 paused for Review; the next Queue reused it and
generated only Chunk 2. The same cached KJ wrapper held eight real preview
images before Continue, while MODEL and sampling signatures remained equal.
- Browser on normal ComfyUI: with the pre-existing audited local KJ profile,
restart and frontend reload succeeded. **Use This Take** and **Continue From
Here** were visibly enabled. Continue From Here followed by the browser Run
button completed Chunk 2: **one reused / one generated**. Core execution times
were approximately 100 and 92 seconds, including decode/save; they are not a
speed comparison. Opening Render History involved one manual click; the
subsequent Continue selection and Run submission were automated browser actions.
- Both GPU pairs preserved the original Chunk 1 raw SHA-256, original Take and
correct parent link. All four 5s/10s video/audio outputs fully decoded. The
upstream GPU pair also passed saved-latent finite-value checks. No execution
errors or OOM occurred. The observed upstream system VRAM maximum was 15,446
MiB at roughly six-second intervals, not an instantaneous per-process peak.
- All 21 existing workflow JSON/ZIP files remain byte-identical. Existing user
Take/workflow files were preserved; there is no storage or schema migration.
These are targeted continuation/Review checks on Core 0.42.1, frontend 1.55.18,
Python 3.13.12, torch 2.14.1+cu130 and RTX 5060 Ti 16GB. Full CPU validation also
uses real Core/provider imports. They do not establish visual/audio quality,
long-video quality, all accelerators, all KJ versions, or save/reload acceptance
of every shipped workflow. The separate Turbo Sample already present on main is
included unchanged; its six-step/4-step-LoRA pairing has no new GPU acceptance.
## 日本語
公開KJ Preview Overrideを接続した新規Runまで「v6 read-only」になる対応漏れを修正し、
再利用不可のTakeには保存時の理由と対処法を表示します。安全判定は緩めていません。
更新後はComfyUIを再起動し、ブラウザーを強制再読込してください。既存Workflowの配線変更や
Run/Take削除は不要です。旧unsafe/v5 Takeは保持されますが、自動的に再利用可能には
なりません。設定修正後は新Runで開始してください。Preview Override側の任意VAE/Audio VAEは
未接続、Tiny VAEはnoneが条件です。通常の生成用VAEは接続したまま使用します。
CPU 2,075成功、公開KJでの実GPU継続、既設KJを維持した実ブラウザーでのボタン有効化と
Continue From Here→Run→Chunk 2完了を確認しました。両GPU構成でChunk 1を保持し、
1再利用・1新規生成、Takeの親子関係、映像・音声全デコードを確認しています。
対象修正の検証結果であり、全KJ構成や画質・長時間品質への包括的な合格ではありません。
+44
View File
@@ -0,0 +1,44 @@
# V3.9 Turbo Sample
A separate, user-supplied example of Continuum prompt formatting, not a new model
or a verified quality preset. Available in Templates as
`MiniMax_H3_Continuum_V39_Turbo_Sample` (no parenthesized suffix).
- Prompt Format: Auto; 2 chunks x 8 seconds; Full Run; Progress OFF.
- The multiline prompt demonstrates `[Chunk 1]`, `[Chunk 2]`, first-image alignment,
continuous dancing with a fixed background/camera, and an off-screen male Thai rap.
`<d>[Thai] ... </d>` is H3 lyric syntax, not part of the template name.
- Saved model: `minimax/DasiwaMinimaxH3_dasiwaHybridV4_int8.safetensors`.
Obtain it from the [author's model page](https://civitai.com/models/2877206?modelVersionId=3391862)
or select a compatible installed H3 model. No model weights are bundled.
- One LoRA is enabled: `minimax/minimax_h3_fl2v_turbo_4step_v1.2_768p_comfyui_bf16.safetensors`,
strength 1, from [LightX2V](https://huggingface.co/lightx2v/Minimax-h3-Turbo).
The supplied graph uses Euler / simple / 6 steps / Draft 0.30 MP. These are
preserved sample values, not a validated 4-step/768p configuration; check the
selected model and LoRA instructions before generation.
- Video VAE: `minimax_h3_video_vae_int8_convrot.safetensors`; audio VAE:
`minimax_h3_audio_vae_fp32.safetensors`; text encoder:
`qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors`, type `minimax`.
See [model download links and folders](../README.md#required-models-and-extensions).
- Select your own First Image (`example.png` is only a placeholder). The disabled
optional audio/video inputs retain sample filenames; select real files before
enabling them. Reselect model/LoRA paths if you do not use a `minimax` subfolder.
- Requires KJNodes (SageAttention and Model Preview Override), rgthree, and Spectrum
node definitions. Spectrum is OFF. Preview Override has video/audio VAE connected;
this is outside the audited v3.9.8 Preview Phase A Take-reuse profile.
- Thai lyric accuracy, unchanged background, identity preservation and subjective
video/audio quality are not established by adding this sample. Thai is outside
H3's explicitly listed stable dialogue languages. No new GPU test is claimed.
- Existing standard workflows and saved Runs/Takes are unchanged. Outputs use the
separate `video/MiniMax_H3_Continuum_V39_Turbo_Sample` prefix. GitHub main includes
this sample; the already-published v3.9.8 release/Registry package is not replaced.
## 日本語
標準Workflowとは別のTurboサンプルです。Auto・2チャンク×8秒で、改行した
映像・環境音・音楽の記述、入力画像の指定、固定背景、画面外の男性タイ語ラップを
示します。画像・モデル・LoRAは各自で導入して選択してください。
4-step/768p LoRAに対して6 steps・Draft解像度という添付の設定を保持しています。
推奨設定や画質の合格を意味しません。タイ語の発音、人物・背景保持も未保証です。
KJ Preview OverrideのVAE接続はv3.9.8の限定Take再利用検証の範囲外です。
今回はサンプル追加とCPU検証のみで、実環境反映・新規GPU試験・Release更新は行いません。
+705
View File
@@ -0,0 +1,705 @@
# H3 Continuum — illustrated user guide
[Back to the quick start](../README.md) · [V3.9 References](V39_REFERENCE_INLINE.md) · [History](README_HISTORY.md)
This guide preserves the detailed operating instructions and screenshots from
the README at source commit `894869d9` (package 3.9.8).
V3.8-labeled controls and screenshots describe the retained V3.8X2 workflow;
the V3.9 setup and Reference sections identify the separate current workflow.
Earlier dated fixes, version statements and measurements are historical evidence,
not a new acceptance claim. Use the [current README](../README.md) and the
[v3.9.8 compatibility profile](KJ_PREVIEW_OVERRIDE_V398.md) for current scope.
- [V3.9 template setup](#start-here-v39-and-the-retained-v38x2-workflow)
- [Reference Images](#v39-reference-images)
- [Complete V3.8 control reference](#complete-v38-ui-reference)
- [Review, retry, Takes and extension](#review-continue-and-revise-a-chunked-video)
<!-- Preserved README sections begin. -->
## Start here: V3.9 and the retained V3.8X2 workflow
The ComfyUI **Templates → ComfyUI-H3-Continuum** gallery offers two separate entries: the user-selected [V3.9 workflow](../examples/workflows/MiniMax_H3_Continuum_V39.json) and [V3.8X2 workflow](../examples/workflows/MiniMax_H3_Continuum_V38X2.json). The [V3.8X2 workflow ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip) contains **both unchanged JSON workflows** for one-download setup; the [V3.9 ZIP](../examples/workflows/MiniMax_H3_Continuum_V39.zip) contains only V3.9. These are workflow archives, not custom-node installers. The V3.8 Sampler remains available, so existing V3.8X2 workflows still load. V3.9 does not automatically convert their Reference connections or saved Runs/Takes: preserve the old workflow and open the dedicated V3.9 graph for new work.
The selected V3.9 template has 32 nodes. Image loaders 1–6 are wired to the new helper and bypassed by default; slots 7–9 remain available but require the user to add and connect their own image loaders. Choose actual files and enable only the desired images. The saved size controls are **First Image + Draft — 0.30 MP**, and the separate First Image loader is enabled; select your own First Image file before Queue. The saved 480×640 Width/Height values apply only in Manual mode. `Reference Image Size` is `Match Output`; it resizes conditioning after source-image load, not the source files themselves. For a clear per-chunk Reference test, turn First Image and Last Image off, switch to Manual, and use Reference Images alone. First/Last additionally constrain the video; Last Image may merge the final two logical chunks into one physical group, which requires the same effective Reference set in both. A subject may remain visible in later chunks through video continuity even after its Reference is no longer assigned.
In the helper, `All chunks` uses connected images throughout; `Per chunk` provides an image-by-chunk table. The template starts in `Per chunk` with all assignments off: after enabling a loader, tick the chunks that should use it. Slot numbers never shift: Image 1 is `@R1` and Image 9 is `@R9`, even when intervening slots are empty. In Sequence Prompt, describe the action for each chunk and use `@R` tags only for active References; their actual `<Picture N>` numbers are resolved per group. Unused tags warn without stopping generation. Start with short chunks and pre-resized images: source images may consume RAM before the Sampler's Reference Image Size setting applies.
V3.9.3 is the current source version on `main`. Download versioned source from [GitHub Releases](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases); ComfyUI Registry publication is separate. Historical Releases and tags remain available. Configured 1024×1024 Reference-only GPU/API runs succeeded, but the unchanged default template and browser save/reload remain separate acceptance gates. For a reproducible older environment, use the matching historical Release/tag and workflow; the retained V3.8 Sampler on current `main` specifically supports V3.8X2 workflows.
`pyproject.toml` declares ComfyUI `>=0.32.0`; earlier real-generation/GPU validation used ComfyUI `0.34.2`, and the scoped Fixed cache comparison above used `0.38.0`. The minimum declaration does not establish V3.9 runtime acceptance on 0.32.0. The previous Windows CPU run is `1696 passed / 1 skipped / 0 failed`; the earlier separate Linux rerun report records `1534 passed / 3 skipped / 0 failed` and does not include the new cache or external INT tests. Neither replaces browser save/reload or GPU acceptance for the unchanged official template.
## V3.9 Reference Images
### Reference Images node: choose the image for each chunk
![H3 Continuum Reference Images V3.9 node with fixed image slots and a per-chunk assignment table](images/v39/v39-reference-images-node.png)
*Example UI, not the downloadable workflow's default assignments.* Image 1–9 are fixed slots: Image 1 is always `@R1`, even when other slots are empty. With `Reference Use = Per chunk`, tick the image/chunk intersections you want; each row's `All` / `Off` buttons select or clear that image across chunks. In this screenshot, Image 1 is assigned to Chunks 1 and 2, while other images have different assignments. The official V3.9 workflow starts with its image loaders bypassed and every assignment unchecked; choose real image files, enable their loaders, then tick the intended chunks.
### Sampler V3.9: connect the helper once
![H3 Continuum Sampler V3.9 with one Reference Images Optional input](images/v39/v39-sampler-node.png)
Connect the helper's green `reference_images` output to the Sampler's single `Reference Images (Optional)` input. The V3.8X2 workflow uses its own V3.8 Sampler and older Reference sockets; opening it does not convert those connections to V3.9. This screenshot shows a **Manual 480×640 example**, not the official V3.9 workflow's saved **First Image + Draft 0.30 MP** setting. `Reference Image Size = Match Output` controls Reference conditioning size after image loading; it does not pre-resize source files or set output resolution.
V3.9 now connects all nine fixed Reference Images through one **H3 Continuum Reference Images V3.9** node. Its in-node table selects All chunks or specific chunks. The V3.9 Sampler has one typed Reference Images connection and keeps Sequence Prompt, First/Last, Review and the rest of its generation controls. Use `@R1` for Image 1; the backend converts active tags to the group's actual `<Picture N>`. Both modes use the same group-routing engine.
Prompt parsing never blocks generation. If an explicit Timeline cannot be parsed, the original text falls back to Fixed with diagnostic `H3C-P100`. An uncovered Timeline chunk reuses the previous prompt and reports `H3C-P101`. Select the image for each chunk in the Reference Images node; `@R1`–`@R9` in Sequence Prompt refer to those fixed image numbers.
For the supplied `2 × 10 s` workflow, put each time header on its own line and describe only that chunk's action. For example, after enabling the matching image loaders and checking their chunk assignments:
```text
[0-10s]
@R1 walks forward through a quiet hallway. The camera follows smoothly.
[10-20s]
@R2 enters from the right. The camera turns to follow @R2.
```
Use the actual subjects and actions from your images; this is a syntax example, not a saved prompt in the downloadable workflow. Do not place prompt text on the same line as `[0-10s]`: that is not a Timeline header. A Reference can still influence a chunk without an explicit `@R` mention when its checkbox is enabled.
The [official V3.9 workflow](../examples/workflows/MiniMax_H3_Continuum_V39.json) is the latest user-supplied JSON, copied without editing; its [single-workflow ZIP](../examples/workflows/MiniMax_H3_Continuum_V39.zip) is also available. It preserves the full V38X2 model/LoRA, Decode Cache, Finalize and Save chain. Six image loaders are prewired (initially bypassed); to use Image 7–9, add `H3 Continuum Load Image` nodes and connect them to the matching helper slots. The Sequence Prompt input remains connected but is intentionally blank: write your own prompt before Queue. The saved graph uses `Timeline`, two 10-second chunks, `Full Run` (displayed as Generate Full Video), First Image + Draft 0.30 MP size selection, and a separate Project ID. The earlier [Reference Inline migration fixture](../examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.json) remains available unchanged. See the [step-by-step Reference Images guide](V39_REFERENCE_INLINE.md). Replace local media and model filenames as needed before Queue; enabling an image requires choosing a real file. Seven to nine active references can be particularly memory-intensive: start with fewer, smaller source images. A full ComfyUI backend restart and browser hard-refresh are required before the new node definitions appear. Browser save/reload and GPU acceptance of this unchanged default graph are still pending; a GitHub Release does not imply these gates or a Registry package have been completed.
**Compatibility:** Existing V3.8X2 workflows still use the V3.8 Sampler and old Reference Images 4–9 helper. V3.9 needs its new helper and connection. Open the new V3.9 workflow instead of only changing the Sampler type in an old graph: older V3.9 prototype workflows with direct 1–3 connections and Sampler-side routing widgets need manual rewiring, and their saved widget positions may differ. Accepted V3.8 Runs/Takes are not automatically migrated. Keep a copy of the previous workflow and use a new Run Name for V3.9.
Earlier browser checks covered the previous V3.9 controls. The new helper has GPU/API execution evidence; its browser save/reload and the unchanged default template's full acceptance are still open. CPU/JS and manifest results are separate; no V3.9 Release/Registry publication is implied.
**Version boundary:** V3.8X2 package 3.8.3 and V3.9.0 are separate workflows. This update does not create a V3.8X2 or V3.9 GitHub Release, tag, maintenance branch, or Registry version; it does not remove any older Release. The existing [`v3.8.0` tag](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/tree/v3.8.0) remains the reproducible V3.8.0 baseline.
## Loader persistence repair on main (2026-09-21, after 3.8.3)
The current paired V3.8X2 workflows now use **Core Load Audio** and **Core Load Video -> H3 Continuum Video Adapter**. The adapter has Force Rate (default 24) and IMAGE/AUDIO outputs, but **no Enable switch or file selector**. Core owns video selection/upload/save; the established duration-preserving frame selection is reused and audio is passed unchanged. To disable Video Guide, bypass the adapter or the whole video-input group; do not bypass only Load Video while leaving its required-input adapter active.
**H3 Continuum Load Image remains**, including Enable Image. Its mode observer now delegates to inherited Core setters (Issue #23), and no longer replaces the widget collection while drawing or saving. Enable is appended after Core's widgets; Core handles the filename and native Bypass persistence. The older Audio/Video node IDs remain loadable as deprecated compatibility nodes, not as the recommended new-workflow loaders. The V3.8X2 baseline registered ten node IDs including these two legacy nodes; V3.9 adds the V3.9 Sampler and Reference Images helper as separate IDs, for twelve total. V3.8 Sampler inputs, generation parameters, Decode Cache, Finalize and existing latent/storage contracts are unchanged.
After updating, restart ComfyUI and reload/refresh the frontend. A workflow that already omitted its filename or saved an incorrect mode cannot recover those missing values: reselect the intended file and Bypass state, then save once. Existing user workflows and saved Takes are not automatically rewritten. Changing upstream node types can legitimately create a new Run Storage revision; preserve the original workflow when resuming accepted runs.
Validation details and limits: [Loader repair record](LOADER_PERSISTENCE_REPAIR.md). CPU/store-model regression tests are not a substitute for the pending Windows Core 0.36.0 / frontend 1.53.6 browser tab-switch acceptance. No Release/tag or Registry publication is implied by this main-branch repair. Older loader screenshots below describe the retained legacy nodes.
## Timeline Video — Experimental on main
The existing video IMAGE input is now displayed as **`Timeline Video Frames`** and has one appended optional control, **`Video Reference Mode`**. Existing workflows and API calls that do not contain the new setting keep the old behavior through the default **`Repeat Reference`** mode.
- **`Repeat Reference`** keeps the legacy Video Guide contract: each physical generation group receives the same bounded prefix from the source. It does not loop or stretch a short source.
- **`Follow Timeline`** uses the 24 fps source interval corresponding to each physical output group. Continuation-prefix frames are not consumed twice. If the source ends, the remaining groups continue without video reference; a final usable partial interval may receive only H3's short frame-grid padding.
This setting is independent of `Prompt Format = Timeline`. It does not add a new socket or loader, does not use the source video's audio automatically, and does not provide frame-perfect motion copying. The complete decoded IMAGE batch is still held in system memory, so this feature is not a long-input RAM optimization.
Use the existing connection:
```text
Core Load Video -> H3 Continuum Video Adapter -> Timeline Video Frames
```
The two comparison workflows are intentionally separate from the unchanged official V3.8X2 workflows:
- [Follow Timeline JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Follow.json) | [ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Follow.zip)
- [Repeat Reference JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Repeat.json) | [ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Repeat.zip)
Local acceptance passed with full CPU `1424 passed / 1 skipped / 0 failed`, browser save/reload, and Follow/Repeat `2 x 5 s` GPU runs at 704x416, 24 fps, 240 frames, and 10-second 32 kHz stereo audio. Follow changed to the later source interval for Chunk 2 while Repeat retained the shared prefix; both completed without OOM, NaN, allocation failure, or crash. Peak observed VRAM was about 15.6–15.7 GiB on the tested RTX 5060 Ti 16 GB system, so 16 GB headroom was small. This is a functional mode-separation result, not a universal speed or subjective-quality guarantee. See [Timeline Video Experimental](TIMELINE_VIDEO_EXPERIMENTAL.md).
> **V3.8X2** is the product and workflow label for package `3.8.3`. It keeps the V3.8 production sampler and adds optional Reference Images 4–9 plus the built-in Decode Cache Helper. The older V3.8.0 package remains available from tag [`v3.8.0`](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/tree/v3.8.0).
<img width="1536" height="1024" alt="exec-9cfa73b4-c1d3-4416-8950-fa661b946638_2" src="https://github.com/user-attachments/assets/8df15764-6db6-4d4c-9eb5-0d2c7b0668f0" />
## Official V3.8X2 workflows
- **Standard name:** [JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2.json) | [ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip)
- **Helper-explicit name:** [JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json) | [ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip)
The two V3.8X2 JSON names contain the **same official graph**; they are distribution aliases, not different configurations. Both include the built-in Decode Cache Helper with its saved `Auto / RAM 256MB / Disk 8GB / reset_token 0` settings and the nine-reference helper connection. The standard V3.8X2 ZIP now also contains the separate V3.9 JSON, while the helper-explicit ZIP contains only its correspondingly named V3.8X2 JSON. Older V3.8X workflows remain available for existing projects.
## Nine Reference Images: memory warning
Reference Images 1–3 connect directly to the Sampler. Optional images 4–9 connect through `H3 Continuum Reference Images` and then to `Reference Images (Optional)`. Using all nine is a high-resource configuration, not the normal minimum requirement.
The acceptance run completed on an **RTX 5060 Ti 16 GB / 64 GB system** only after the nine source references were kept near **0.30 MP each**. With `512 × 608` output, `3 × 5 seconds`, Spectrum Off, LoRA Off, Sage Attention, and Balanced 22-frame continuity, Sampling used about **15.0–15.5 GiB of 16,311 MiB VRAM**. There was little 16 GB headroom.
When using many references, especially all nine, resize copies of the source images to about 0.30 MP before loading or connecting them. The Sampler still performs its normal conditioning resize, but smaller source files reduce input and preprocessing pressure. Larger references, larger output sizes, extra wrappers, or concurrent GPU work may require a higher-spec GPU and more system RAM. Nine references are optional and are not guaranteed to improve quality over three.
## Built-in Decode Cache Helper
### Important: this is not an unconditional new-generation speedup
**Manual cache clear:** normally no action is needed. Press the Helper's "キャッシュをクリア" (Clear cache) button to discard that Helper's cache on the next Queue and run native Decode again. It does not delete immediately or queue automatically. The internal `reset_token` remains serialized for workflow/API compatibility and increments once per click (wrapping to 0 at the INT limit). Creating, loading, or cloning a node does not change it. Without JavaScript, or in API workflows, change `reset_token` through the traditional numeric input.
**The Helper saves time only when unchanged latents are decoded again. It does not accelerate Sampling.** A first MISS runs native Decode and stores the result; hashing and storage can make that run slower. Generating different latents with a new random seed each time normally produces MISS, so this feature should not be expected to speed up that usage.
Concrete reuse scenarios:
- **Continue by adding chunks:** when existing chunk latents are retained and decoded again for output, cached existing chunks can HIT; newly generated chunks initially MISS.
- **Fix / partial regeneration:** unchanged chunk latents can HIT when submitted again; regenerated chunks whose latent contents changed MISS.
- **Complete / Resume output:** decoding retained latents again without resampling can benefit. Merely playing an existing video file is not a cache use case.
- **Retry from Decode onward, or change Finalize/output settings:** reuse applies when the same latents and decode-related metadata reach the Helper and Decode runs again.
- **Regenerate with a fixed seed:** reuse requires identical resulting latent contents. The same seed alone does not guarantee HIT.
The action names themselves do not enable reuse. **Each Video/Audio entry must match in latent contents, shape, dtype, decode-related metadata, and VAE identity, and its cache entry must still exist in that Helper.** If ComfyUI reuses an existing output without executing Decode, there is no additional Decode saving attributable to this Helper. Connect the Helper and select `Auto` or `RAM`. In its report, `hit` means reuse, `miss` means native Decode, and `off` / `bypass` mean no cache reuse.
Cache reuse is process-local and does not survive Python restarts. Eviction and VAE identity changes can also cause MISS. Changing mode, RAM/Disk budgets, or `reset_token` clears this implementation's Helper cache. Keeping `reset_token = 0` does not reset it on every Queue.
V3.8X2 ships `H3DecodeCacheHelper` inside this Continuum package. It remains a separate public node under `MiniMax H3/Continuum/Helpers`; no second custom-node addon is needed. The original nine IDs remain registered for compatibility; the loader migration adds H3ContinuumVideoAdapter. Sampler, Finalize, Assembly Plan, Core Decode, and saved Sampler widget/socket contracts are unchanged.
Connect Sampler `video_latents` and `audio_latents` plus their native Video/Audio VAE inputs to the Helper, then connect Helper `images` and `audio` to Finalize. Keep the Sampler `assembly_plan` directly connected to Finalize. The [official V3.8X2 workflow](../examples/workflows/MiniMax_H3_Continuum_V38X2.json) preserves its saved graph and settings. It contains no Core VAE Decode nodes, so returning to Core direct Decode requires adding those two Core nodes and reconnecting Finalize, or loading a prior Core-direct V3.8X workflow. Do not simply remove the Helper and expect automatic rewiring.
`Auto` caches full physical Video decodes on private process-local disk and small Audio decodes in bounded RAM. `Off` delegates native Decode and clears the Helper cache. Cache failure falls back to native Decode; native Decode, OOM and queue interruption are not retried. Files are not reused across Python restarts. Keep only one provider of the `H3DecodeCacheHelper` node ID: the previous standalone addon must be disabled or removed during the migration, after a verified built-in installation.
At 736×416 (0.306MP), 3×5 seconds, Euler/simple 6 steps, Turbo FL2V v1.2, a warm backend and three Video HITs, one measured configuration produced a **mechanistic estimate of about 47 seconds / 24.5% less Full Run Total**. One direct A/B pair observed 50.887 seconds / 26.46%, and two Complete/Resume re-decode pairs averaged 46.309 seconds / 79.826% less Total. These are narrow observations, not repeated universal averages or an unconditional 20%+ speed promise. The first MISS can be slower, new Sampling is not accelerated, and a changed latent or VAE identity produces a MISS.
**Historical V3.8X workflow: [JSON](../examples/workflows/MiniMax_H3_Continuum_V38x.json) | [ZIP](../examples/workflows/MiniMax_H3_Continuum_V38x.zip)** — [Published releases](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases)
✅ V3.8X on `main` includes the September 8, 2026 Review hotfix: Review Each Chunk continuation, stale `Regenerate From` state, completed-sequence extension, and Render History queue handling are repaired. Update with `git pull --ff-only origin main` (or ComfyUI Manager **Update**), restart ComfyUI, and hard-refresh the browser if the old UI remains. Existing saved Takes are preserved. Issue #13 remains a separate open long-continuation quality issue and is not part of this hotfix.
H3 Continuum is a Production Sampler for generating, reviewing, partially regenerating, and resuming long-form MiniMax H3 video without restarting the entire work. V3.8 has two product layers: **Main / Production** and **Advanced**.
## Install or update first
For a new Git installation:
```bash
cd ComfyUI/custom_nodes
git clone https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum.git
```
For an existing Git checkout:
```bash
cd ComfyUI/custom_nodes/ComfyUI-H3-Continuum
git pull --ff-only origin main
```
Restart ComfyUI after cloning or pulling. If ComfyUI Manager installed the node, use its **Update** action instead of mixing Manager updates with a second Git checkout.
To install the historical V3.8.0 package into a separate checkout instead:
```bash
cd ComfyUI/custom_nodes
git clone --branch v3.8.0 --single-branch https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum.git ComfyUI-H3-Continuum-v3.8.0
```
Use only one Continuum checkout at a time. ComfyUI Manager **Update** tracks `main`; it does not select historical tags.
Download the current V3.8X2 workflow: [JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2.json) or [combined V3.8X2/V3.9 ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip). The V3.8X2 JSON is one Spectrum-capable graph, also usable with [LightX2V Turbo](https://github.com/ModelTC/Minimax-H3-Turbo); both Spectrum and all Turbo LoRA entries are saved disabled. Install its external Spectrum, rgthree, and KJNodes nodes before opening it. ComfyUI-Easy-Use is not required. See [Spectrum and Turbo setup](#turbo-lora-and-spectrum-in-supplied-workflows) below.
## Prompt and skill downloads
- [V3.9 Reference Images prompt skill](../write-h3-v39-reference-prompts/SKILL.md) | [ZIP](../write-h3-v39-reference-prompts.zip) — for fixed `@R1`–`@R9` image slots and per-chunk assignments. Use it with the V3.9 Reference Images node; it is separate from the general Continuum prompt skill below.
- [LLM system prompt ZIP](../H3-Continuum-LLM-System-Prompt-v1.zip) — system instructions and reference material for Continuum prompt authoring.
- [Continuum prompt skill ZIP](../H3-Continuum-Skill-v1.zip) — general chunk-aware prompt authoring for Codex and other compatible agents.
- [Continuum Dance Director skill ZIP](../minimax-h3-continuum-dance-director.zip) — long-form choreography prompts for one primary dancer, with body and camera continuity across chunks.
These are optional prompt-authoring resources, not ComfyUI custom nodes. Extract each ZIP and follow its included instructions. They do not change the sampler or guarantee generated motion or image quality.
## You can ask an AI to read this manual
This README is intentionally detailed. You do not have to memorize it. Give its GitHub URL or this file to a web-enabled AI, describe your target video, GPU/VRAM, available First Image/audio, and whether you want to review every chunk. Ask it to use the **exact V3.8 labels** below.
Example request:
```text
Read the current ComfyUI-H3-Continuum V3.8 README. I have a 16 GB GPU and want
to create a 30-second I2VA video, reviewing every chunk. Tell me exactly which
Continuum inputs and controls to use, what must stay fixed between Queue runs,
and what output length I should expect after each action. Do not use older V3.7 UI.
```
AI products, plans, and repository-reading behavior change frequently. If a URL is not read correctly, paste or upload this README instead. Do not send private workflows, local paths, tokens, or credentials to a public AI service. The README and the UI remain the source of truth; an AI summary can still be wrong.
## V3.8 supported surface
The V3.8X2 baseline registers ten node IDs, including two deprecated loader compatibility IDs; V3.9 adds a separate Sampler and Reference Images helper. Current official workflows use Core Load Audio and Core Load Video plus the Video Adapter.
- **H3 Continuum Sampler V3.8** — the Main sampler
- **H3 Continuum Finalize** — decoded Video/Audio assembly with optional seam handling
- **H3 Continuum Load Image**
- **H3 Continuum Load Audio** (legacy compatibility; new workflows use Core Load Audio)
- **H3 Continuum Load Video** (legacy compatibility; new workflows use Core Load Video + Video Adapter)
- **H3 Continuum Video Adapter** — Core VIDEO to H3 IMAGE/AUDIO, default 24fps, no Enable
- **H3 Continuum Second Pass** — the Advanced bridge for external latent processing or upscaling
- **H3 Continuum Reference Audios** — an ordered Reference Audio 1/2/3 bundle helper
- **H3 Continuum Reference Images** — optional additional Reference Images 4–9
- **Decode Cache Helper** — the existing optional decoded-result cache
The current frontend uses `Show Advanced Settings` / `Hide Advanced Settings` to change presentation without changing generation values. Saved drafts that still contain the former frontend-only `H3 Continuum View` property are migrated to the matching collapsed/expanded view and the legacy property is removed. If the frontend extension is unavailable, the complete Python-defined interface remains visible and executable.
The official V3.8X2 workflow uses the Helper path:
```text
H3 Continuum Sampler V3.8
-> Decode Cache Helper
-> H3 Continuum Finalize
-> Create Video
-> Save Video
```
The Helper delegates cache MISS entries to native Core Video/Audio Decode. A manually wired direct Core Decode path remains supported, but it has no decoded-result cache. Save and upscaling remain external. Hi-Res Fix is not part of the V3.8X2 standard workflow; use Second Pass as the bridge when an external latent processor or upscaler is needed. External processors may change only Video LATENT spatial geometry while preserving physical groups, B/C/T, finite values, first-pass Audio, and the original Assembly Plan. Finalize accepts public `IMAGE`/`AUDIO` outputs and does not require a specific decoder class. SageAttention, Sol-Attn, and Spectrum remain external MODEL wrappers. See the [V3.8 Open Integration Contract](V38_OPEN_INTEGRATION_CONTRACT.md).
## Complete V3.8 UI reference
This section covers every user-facing control and socket on the seven public V3.8 nodes. Names such as `Power Lora Loader (rgthree)`, Spectrum, SageAttention, Core VAE Decode, and Core Save Video belong to ComfyUI or another extension. They may be used in a supplied workflow, but they are not Continuum controls.
### H3 Continuum Sampler V3.8: required graph inputs
| Input | What it does | Normal use |
|---|---|---|
| `model` | MiniMax H3 diffusion model, including any upstream MODEL wrappers or selected LoRA | Required |
| `clip` | MiniMax H3 text encoder | Required |
| `video_vae` | Encodes connected image conditioning; Continuum does not use it for final decoding | Required by the node; unused by pure T2VA |
| `sampler` | ComfyUI sampling algorithm | Required |
| `sigmas` | Noise schedule and effective step count | Required |
| `Sequence Prompt` | Complete text, list, timeline, or JSON sequence prompt | Required |
The Sampler returns six outputs: a list of `video_latents`, a list of `audio_latents`, one `assembly_plan`, a text `status`, the selected `driving_audio` when used, and a `refine_context` for the Advanced Second Pass. Raw latent lists normally go through Core Video/Audio Decode before Finalize.
### H3 Continuum Sampler V3.8: optional media inputs
| Input | Role | Important distinction |
|---|---|---|
| `first_frame` | First Image for I2VA or FL2VA | Can also supply the output aspect ratio when `Size Source = First Image` |
| `last_frame` | Optional final-image constraint for FL2VA | Can trigger a Long Terminal Merge; the final pair may become one atomic review unit |
| `reference_image_1`–`reference_image_3` | Ordered appearance, identity, subject, or scene references | They never become the implicit size source |
| `Reference Images (Optional)` | Additional images 4–9 bundled by `H3 Continuum Reference Images` | Empty inputs are skipped; encoding remains in the Sampler |
| `Video Guide Frames` | A video loader's IMAGE frame batch, applied as a persistent guide | It does not carry the source video's audio; frames are interpreted at 24 fps |
| `Driving Audio` + `Driving Audio VAE` | Original audio timeline used as native guide conditioning | The selected source audio becomes final audio; generated audio and Audio Seam are bypassed |
| `Reference Audio (Optional)` + `Reference Audio VAE (Optional)` | Legacy single conditioning-only audio reference | Generated audio remains final audio |
| `Audio References (Optional)` | Ordered bundle from `H3 Continuum Reference Audios` | Do not connect this together with the legacy single Reference Audio path |
| `Still Image Guide (Optional)` | Compatibility socket inherited from the V3.7 guide contract | Experimental; not part of the V3.8X2 official workflow |
#### First Image, Last Image, and Reference Images
![First Image enabled and Last Image bypassed](images/v38-manual/first-last-image-bypass.png)
- **First Image** establishes the opening image and is the normal visual starting point for I2VA/FL2VA. It is also the only media input that can drive `Size Source = First Image`.
- **Last Image** constrains the sequence ending. Leave it `OFF` for T2VA and ordinary I2VA. With a connected Last Image, extending or regenerating the sequence may rebuild the terminal pair.
- **Reference Images 1–9** guide identity or appearance throughout generation. Inputs 1–3 remain on the Sampler; optional inputs 4–9 are on `H3 Continuum Reference Images`, connected to the Sampler's `Reference Images (Optional)` socket. The helper only bundles images: resizing, hashing, and VAE encoding remain in the existing Sampler path. Empty slots are skipped; Picture numbers follow active slot order after any First/Last Image. Existing three-loader templates still work unchanged.
Saved workflows with the earlier direct image 4/5 inputs migrate those links into a bundle after graph loading. Migration preserves the original source links and only removes a legacy input after its new link is verified. If a source is unavailable or its destination is occupied, the legacy link is retained with a warning. Old API prompts remain accepted; supplying both legacy and bundle images for the same slot is an explicit input conflict.
![Three optional Reference Image loaders](images/v38-manual/reference-images-three-bypass.png)
Every `H3 Continuum Load Image` has `Enable Image`. `ON` loads through ComfyUI Core. `OFF` uses native node bypass and makes that optional path behave as unconnected without deleting the node or cable. First, Last, and up to nine Reference Image inputs can therefore be connected without forcing every input to be active. An empty image bundle returns no reference and does not change the zero-to-three-image sampling/reuse contract.
| Public loader | Controls | Outputs |
|---|---|---|
| `H3 Continuum Load Image` | `Enable Image`, `image` file | `IMAGE`, `MASK` |
| `H3 Continuum Load Audio` | `Enable Audio`, `audio` file | `AUDIO` |
| `H3 Continuum Load Video` | `Enable Video`, `Video` file, `Force Rate` | `images`, `audio` |
#### Video, audio, and their bypass switches
![Continuum Video and Audio loaders with Enable controls](images/v38-manual/input-video-audio-bypass.png)
- `H3 Continuum Load Video`: `Enable Video`, `Video`, and `Force Rate`. `Force Rate = 0` uses source FPS; a positive value drops/duplicates frames to that rate while preserving nominal duration and audio. Connect `images` to `Video Guide Frames` when visual video guidance is wanted. Connect `audio` to `Driving Audio` only when the source audio should guide and become the final output.
- `H3 Continuum Load Audio`: `Enable Audio` and the audio file. Use it for Driving Audio or Reference Audio according to the socket you connect.
- The node titled `Audio Switch` in the example image is ComfyUI Core's `If/Else Switch` (`ComfySwitchNode`); `Fast Groups Bypasser (rgthree)` belongs to the rgthree extension. Neither is a Continuum node or a requirement of Continuum. The Continuum loaders already provide their own native ON/OFF controls.
`Audio Continuity` is different from both audio inputs: it passes the previously generated audio context into the next generated chunk. It does not import a source track. Driving Audio replaces final generated audio; Reference Audio only conditions generation.
### Main Sampler controls
| Displayed control | Meaning | Recommended starting point |
|---|---|---|
| `Prompt Format` | `Auto`, `Fixed`, `List`, or `Timeline` interpretation of `Sequence Prompt` | `Auto` unless you need to force one parser |
| `Continuity` | Prior Video context retained at each chunk boundary | `Balanced — 22 frames` for Standard audiovisual continuation |
| `Base Seed` | Base for deterministic per-chunk seed derivation | Keep the value unchanged when continuing, comparing, or selecting Takes |
| `Control After Generate` | What ComfyUI does to `Base Seed` after a Queue | `fixed` for `Review Each Chunk`, resume, and controlled comparisons |
| `Audio Continuity` | Carries generated-audio context across boundaries | `true` for normal audiovisual generation |
| `Chunks` | Planned total chunk count, from 1 to 16 | Enter the final total, not the number to add next |
| `Seconds per Chunk` | Duration shared by every chunk | 5–15 seconds is the validated range; see duration notes below |
| `Total Length` | Read-only `Chunks × Seconds per Chunk` | Check only; it never edits either input |
| `Size Source` | `First Image` or `Manual` output geometry | First Image for I2VA/FL2VA; Manual for T2VA or exact dimensions |
| `Resolution` | First-Image sizing preset | `Draft — 0.30 MP` for tests, `Balanced — 0.60 MP` for more detail, or `Native 768` for the H3-native short edge |
| `Custom MP` | Custom First-Image pixel budget | Visible only when `Resolution = Custom` |
| `Width` / `Height` | Exact output canvas in 32-pixel steps | Visible/editable only in Manual mode; 32–16384 is accepted |
| `Run` | `Generate Full Video` or `Review Each Chunk` | Use Review when you want a human decision after each physical group |
| `Progress` | Saved raw chunks, safe resume, and Take history | Review turns it On automatically; enable it for resume/regeneration |
| `Ready to Queue` | Plain-language summary of the next Queue | Read it before pressing ComfyUI's top-right blue `Run` button |
| `Show Advanced Settings` / `Hide Advanced Settings` | Shows or hides technical controls | Visibility only; it does not change their values |
`Continuity = Balanced — 22 frames` with `Continuation Backend = Standard` and `Audio Continuity = true` uses the current Masked AV route. `Fast — 5 frames`, `Strong — 39 frames (Experimental)`, and `Auto — conservative` safely resolve to the older Reference Context route while generated-audio continuity is on. With Audio Continuity off, Standard uses the Video-only masked route.
#### Choose First Image sizing or Manual sizing
![First Image sizing with Draft 0.30 MP](images/v38-manual/sampler-size-first-image-draft.png)
`Size Source = First Image` preserves the connected First Image aspect ratio. `Resolution` then chooses its pixel area:
| Preset | Target |
|---|---|
| `Draft — 0.30 MP` | Fastest and lowest-memory First Image starting point |
| `Balanced — 0.60 MP` | More detail with greater memory and processing cost |
| `Native 768` | 768 px short-edge target with a 1344 px long-edge cap |
| `Custom` | `Custom MP` selected by the user; separate from the Native 768 cap |
![Size Source First Image tooltip](images/v38-manual/size-source-first-image-tooltip.png)
`Size Source = Manual` ignores the preset and uses exact Width/Height values. Both must be multiples of 32. Manual is the normal choice for T2VA and also supports a 32×32 diagnostic canvas when only generated audio is being checked.
![Manual Size Source with explicit Width and Height](images/v38-manual/size-source-manual-640.png)
If First Image mode is selected but no usable First Image reaches the Sampler, V3.8 uses the stored Manual Width/Height as a safe fallback and reports the dimensions in `status`. Select `Size Source = Manual` to inspect or edit those values; Width/Height are hidden in the First Image facade. Reference Images and Video Guide never become implicit size sources. The historical `Auto / Landscape / Portrait / Square` values are migration/API compatibility values, not current Main choices.
Size selection is built into the V3.8 Sampler. A separate megapixel, empty-latent, or image-size node is not required for the normal Continuum workflow.
#### Choose a chunk duration
- **5 seconds** is a valid H3 duration and the fastest practical choice for workflow checks, Review UI checks, and quick Takes.
- **8–10 seconds** usually reduces the number of handoffs while keeping each Queue easier to review and rerun.
- **15 seconds** is the [upstream MiniMax H3 native maximum per generation](https://github.com/MiniMax-AI/cli/blob/main/skill/h3-video/SKILL.md) and minimizes the number of boundaries, but it increases per-chunk time and memory. Longer is not automatically higher quality.
- Continuum accepts 4–30 seconds for compatibility and experimentation. Values above 15 seconds are not the same as an upstream native-duration recommendation and can be expensive at high resolution.
For an audio-only listening diagnostic, use T2VA, `Size Source = Manual`, and a very small canvas such as 32×32, then play the saved video and ignore its picture. This is useful for hearing music or boundary continuity; it is not evidence of normal-resolution video quality.
### Advanced Sampler controls
| Control | When it appears | Meaning |
|---|---|---|
| `Continuation Backend` | Advanced open | `Standard` is the V3.8 path; `Compatibility` restores the older Reference Context route for comparison |
| `Regenerate From` | Advanced open and Progress On | `Auto` resumes the longest compatible prefix; `Chunk N` reuses earlier chunks and rebuilds N through the target |
| `Variation Nonce` | Explicit `Regenerate From = Chunk N` | `0`: automatic variation selection with interrupted-run resume; `1` or higher: a fixed variation value. Base Seed and the other generation settings stay unchanged |
| `Run Name (Optional Override)` | Advanced open and Progress On | Stable name for resume and Render History; blank uses the Sampler's automatic identity |
| `Reference Image Size` | Advanced open and a Reference Image connected | `Match Output` is practical; `Max Identity` preserves more reference detail and may use more memory |
| `Video Guide Size` | Advanced open and Video Guide connected | `Efficient — 0.4 MP`, `Balanced — 0.6 MP`, or `Match Output` |
Frontend-managed IDs, selected Take IDs, one-shot review actions, diagnostics, preview, and legacy compatibility values are deliberately not editable as ordinary node widgets. The corresponding settings or action buttons below are the supported interface.
### Review-state controls
When `Run = Review Each Chunk`, `Progress` is On, and a review unit is ready, the settings view changes to this action view:
| Button | What the next top-right ComfyUI `Run` does |
|---|---|
| `Use it and continue` | Accepts the current result, reuses it, and generates one next physical group |
| `Try this chunk again` | Keeps the earlier accepted prefix and creates another Take of the current review unit |
| `Start again from Chunk 1` | Selects a fresh branch from Chunk 1 with current settings and automatic variation; press Queue afterward. Saved Takes are kept. The selection resets to Auto after Queue |
| `Use it and finish the rest` | Accepts the current result and generates every remaining group without further review pauses |
| `Back to Settings` | Shows the normal settings; does not queue, reset, or discard the review |
| `Return to Review` | Returns from settings to the pending review; does not undo edited values |
| `Render History — N Takes` | Opens or closes stored Take browsing; N is Take count, not chunk count |
| `Previous Take` / `Next Take` | Changes the selected stored Take only |
| `Use This Take` | Makes the selected Take canonical on the next Queue; no new sampling is required |
| `Continue From Here` | Branches from the selected Take and generates the next review unit |
The action button selects intent and gains a `✓`; it does not start generation. Press ComfyUI's top-right blue `Run` button afterward. Full step-by-step examples begin in [Review, continue, and revise a chunked video](#review-continue-and-revise-a-chunked-video).
### H3 Continuum Reference Audios
`H3 Continuum Reference Audios` bundles up to three standalone audio references behind one `Audio References (Optional)` Sampler socket. Connect the references without gaps and use `<Audio 1>`, `<Audio 2>`, and `<Audio 3>` in the same order in the prompt. One shared Core Audio VAE encodes each reference independently. This path never replaces generated final audio and does not change the separate Driving Audio contract. Existing workflows may keep using the legacy single `Reference Audio (Optional)` socket; do not connect the legacy and bundled paths together.
Its complete interface is `Reference Audio 1 (Optional)`, `Reference Audio 2 (Optional)`, `Reference Audio 3 (Optional)`, one shared `Reference Audio VAE`, and the `audio_references` output. Leave no gap in the connected order: do not connect Audio 1 and Audio 3 while leaving Audio 2 empty.
The small arrow/triangle seen beside `audio_references` is ComfyUI's connection direction/cable rendering, not an additional option. Place this helper to the left of the Sampler or use a reroute node if the cable crosses the node body.
### H3 Continuum Finalize
Finalize is normally left wired in the workflow rather than operated repeatedly. It assembles the decoded physical groups into the exact requested timeline. Required inputs are decoded `images`, decoded `audio`, and the Sampler's `assembly_plan`; optional `driving_audio` preserves the selected source track.
| Control | Meaning | Normal choice |
|---|---|---|
| `Audio Seam` | `Auto` corrects decoded audio boundaries only; `Off` leaves them unchanged | `Auto` unless diagnosing audio |
| `Video Seam` | `Auto`, experimental `Auto 2`, `Analyze Only`, or `Off` | Keep the supplied workflow value; use Analyze Only when measuring without altering frames |
| `Buffer Backend` | `Auto`, `RAM`, or `Disk-backed` storage for the assembled IMAGE buffer | `Auto` |
`Exact Total Duration` is managed as On and `Report Detail` is controlled from ComfyUI Settings. Driving Audio bypasses generated audio and Audio Seam, but Video Seam can still apply. Finalize outputs final `images`, final `audio`, and a text `report` for the downstream Save Video node.
### H3 Continuum Second Pass
Second Pass is the Advanced bridge after the complete first pass. It takes `model`, `clip`, `sampler`, low-denoise `sigmas`, externally processed `video_latents`, the matching first-pass `audio_latents`, the original `assembly_plan`, and a separate `refine_seed`. Optional `refine_context` restores the physical group's First/Last/Reference conditioning; optional `video_vae` is used only when that context requires image re-encoding.
Its outputs are `refined_video_latents`, bit-exact first-pass `audio_latents`, an `updated_assembly_plan`, and `status`. Complete Review first: Second Pass is not supported on a partial Review sequence. It is not an automatic Hi-Res button and does not include an upscaler.
### ComfyUI Settings added by Continuum
| Setting | Default | Effect |
|---|---:|---|
| `H3 Continuum: Sampling Preview` | On | Shows live sampling previews; Off only reduces preview overhead |
| `H3 Continuum: Developer Diagnostics` | Off | Enables developer-only logging/assertions; leave Off for normal use |
| `H3 Continuum: Detailed Report` | Off | Adds read-only detail to Sampler status and Finalize report; generated tensors are unchanged |
### Turbo LoRA and Spectrum in supplied workflows
![Example third-party LoRA selector with one LightX2V Turbo LoRA enabled](images/v38-manual/turbo-lora-choices.png)
The four MiniMax H3 Turbo files shown in the example come from the LightX2V MiniMax-H3-Turbo project: [GitHub](https://github.com/ModelTC/Minimax-H3-Turbo) and [Hugging Face files](https://huggingface.co/lightx2v/Minimax-h3-Turbo). `FL2VA` files are for T2VA/I2VA/first-last workflows; `Ref2VA` is for Reference-to-Video. `_768p` identifies the training-resolution family. Select **one LoRA matching the task**—do not enable all four together merely because they are listed.
| File shown | Intended path | Distilled target recorded by the file/repository |
|---|---|---|
| `minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors` | FL2VA / T2VA, 544p family | 8-step; the upstream table also lists 4-step inference as an option for this checkpoint |
| `minimax_h3_fl2v_turbo_8step_v1.0_768p_comfyui_bf16.safetensors` | FL2VA / T2VA, 768p family | 8-step |
| `minimax_h3_ref2v_turbo_8step_v1.0_768p_comfyui_bf16.safetensors` | Ref2VA, 768p family | 8-step |
| `minimax_h3_fl2v_turbo_4step_v1.2_768p_comfyui_bf16.safetensors` | FL2VA / T2VA, 768p family | 4-step |
The number in a Turbo filename is its distilled NFE target. Use the matching upstream recommendation unless a provided workflow explicitly documents a tested exception. In particular, using a 4-step LoRA with `Steps = 6` is a workflow-specific experiment, not the LightX2V default. `Euler`/`simple`, the actual Steps value, and the selected LoRA must be recorded together when comparing results.
`Power Lora Loader (rgthree)`, KJNodes SageAttention, and Spectrum are external components. Continuum does not install, enable, or tune them. The [V3.8X2 workflow](../examples/workflows/MiniMax_H3_Continuum_V38X2.json) is the supplied graph, distributed under two official filename aliases as matching JSON and ZIP pairs. Its saved default has Spectrum disabled and no Turbo LoRA enabled. These three custom-node packages are required to open the complete graph, including when switching it to Turbo. ComfyUI-Easy-Use is not required.
The supplied prompt, media selections, and node titles have deliberately been preserved. Choose files available on your computer, disable unused optional inputs, and enter your own prompt before generating. Models, LoRAs, images, and audio are not included. A custom title such as `Save 3x5s Video` is only a saved label: actual duration follows the Sampler's `Chunks` and `Seconds per Chunk`, not that title.
The same graph can be used as the quick Turbo path: disable Spectrum, enable exactly one task-matched LightX2V Turbo LoRA, then select the sampler and Steps intended for that LoRA or for a separately documented tested variation. The supplied graph lists the FL2VA Turbo 4-step v1.2 768p file but leaves it disabled in the Spectrum default. The previously tested `Euler` / `simple` / `Steps = 6` setup with that 4-step LoRA is a workflow-specific variation, not the LightX2V upstream default. Do not enable Spectrum and a Turbo LoRA together merely because both controls are present.
### Known long-continuation limitation: Issue #13
[Issue #13](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/issues/13) remains open. Repeated continuation can increase contrast, edge energy, or apparent sharpness on some prompts and trajectories. V3.8 Standard has **no active production mitigation** for this issue; it preserves the accepted Masked AV behavior.
An experimental prefix-amplitude renormalization produced a strong improvement in one `6 × 5 s` run, but a later frozen `4 × 8 s` A/B/C gate did not reproduce the target drift and therefore applied gains of `1.0`; Standard and the experimental arm were bit-exact. The older Compatibility route did not justify replacing Standard and raised a separate boundary-audio concern. The accurate release statement is therefore: **the issue is not proven fixed, the current Standard path is unchanged, and experimental mitigation stays Default Off**. Report the first and worst affected chunk, workflow JSON, prompt, seed, model/LoRA, Steps/SIGMAS, size, Continuity, and Audio Continuity when reproducing it.
> **Support boundary:** The original V3.8 launch surface had seven IDs. V3.8X2 registers ten IDs including the Video Adapter and two deprecated loader compatibility IDs; V3.9 adds two more, for twelve. Some retain earlier IDs, including Finalize (`H3ContinuumAssembleSeamV35`) and Second Pass (`H3ContinuumSecondPassV35`); this does not make all older workflows compatible. Use the matching historical Release/tag for unsupported node IDs. See [V3.8 Release and Migration Policy](V38_RELEASE_AND_MIGRATION.md).
The accepted 16GB GPU gates reached about `15.5-15.6 GiB` in the most complex cases. Actual use varies by GPU, driver, backend, model precision, resolution, and connected nodes; this is not a universal 16GB guarantee.
### Review, continue, and revise a chunked video
> **Live UI verification:** The repaired Review flow was rechecked in ComfyUI on 2026-09-07. The current walkthrough used Turbo4, actual `Steps = 4`, Euler/simple, `Draft — 0.30 MP`, and Spectrum Off. Chunk 1 reached the three-action review screen; `Use it and continue` then reused the accepted first chunk, generated only Chunk 2, and reached `Saved sequence is complete`. `Try this chunk again` was also live-validated after the Review repair. Timings below are measurements from that machine, not speed guarantees.
Start here when you already have a working V3.8 workflow and want to keep good chunks, retry a result, or extend a sequence. These instructions use the labels displayed on **H3 Continuum Sampler V3.8**. The **blue `Run` button at the top right of ComfyUI** starts execution. It is different from the Sampler's `Run` setting (`Review Each Chunk` or `Generate Full Video`). Continuum's review messages still say “press Queue”; on the frontend shown here, that means clicking the top-right blue `Run` button. Older frontends may label that button `Queue`.
![ComfyUI's top-right blue Run button starts the selected action](images/v38-manual/live-execute-button.png)
Choose the task you need:
- [Prepare the settings for continuation](#1-prepare-the-settings-for-continuation)
- [Understand the first, shorter output](#2-generate-and-review-the-first-chunk)
- [Keep the result and generate the next chunk](#3-keep-the-result-and-generate-the-next-chunk)
- [Try the current chunk again](#4-try-the-current-chunk-again)
- [Keep the result and finish everything remaining](#5-keep-the-result-and-finish-everything-remaining)
- [Go back to settings without losing the review](#6-go-back-to-settings-without-losing-the-review)
- [Choose an earlier Take or branch from it](#7-choose-an-earlier-take-or-branch-from-it)
- [Regenerate from a particular chunk](#8-regenerate-from-a-particular-chunk)
- [Add more chunks to a completed sequence](#9-add-more-chunks-to-a-completed-sequence)
- [Understand continuation time and check reuse](#10-understand-continuation-time-and-check-reuse)
- [Troubleshoot a missing action or unexpected result](#11-troubleshoot-a-missing-action-or-unexpected-result)
#### Before you begin: total length is not the next Queue's output length
`Chunks` is the **total number of chunks you want**, not the number to add on the next Queue. `Seconds per Chunk` is the duration of each chunk. `Total Length` is the planned final duration; it is not a progress counter.
For example, `Chunks = 6` and `Seconds per Chunk = 5` means a 30-second target. With `Run = Review Each Chunk`, the first Queue normally creates only Chunk 1: a 5-second video. Continuing generates the next review unit. During review, the preview/save output is the **current review unit** (normally 5 seconds), not an accumulating preview. After the run reaches `complete`, or after `Use it and finish the rest` generates the remaining chunks, Finalize outputs the assembled completed sequence.
The screenshots show **two chunks as a small example**. The same controls apply to three, four, five, six, or another supported total.
> **Last Image exception:** The examples below assume ordinary, separately generated chunks. When Last Image uses Long Terminal Merge, the final pair is generated and reviewed together. Continuing or retrying that pair operates on both chunks as one unit. Do not expect a separate review stop between them.
#### 1. Prepare the settings for continuation
1. Find `Base Seed` and keep its current number. You do not need to copy the number in the screenshot.
2. Directly below it, set `Control After Generate` to `fixed`.
![Base Seed and Control After Generate set to fixed](images/v38-manual/seed-fixed.png)
3. Set `Chunks` to your intended **final total**. Use `2` for the screenshot's example, or your own total such as `6`.
4. Set `Seconds per Chunk` to the duration you want. The example uses `5`.
5. Check `Total Length`: two chunks of 5 seconds should show `10 seconds`; six should show `30 seconds`.
6. Set `Run` to `Review Each Chunk`.
7. Set `Progress` to `On — Resume and Takes available`.
8. If `Regenerate From` is visible under `Show Advanced Settings`, leave it at `Auto` for normal continuation.
9. Check that the green `Ready to Queue` card describes the total and says `Review each chunk`.
![Review settings with fixed seed, two 5-second chunks, Review Each Chunk, and Progress On](images/v38-manual/live-review-settings.png)
**Keep these settings while continuing:** `Base Seed`, `Control After Generate = fixed`, `Seconds per Chunk`, and `Progress = On — Resume and Takes available`. Keep the same workflow, prompt, input media, model, sampler, step schedule, size, and continuation settings too. Leave `Run Name (Optional Override)` unchanged. Changing generation inputs can make saved chunks incompatible; keeping the seed alone does not guarantee reuse.
For ordinary continuation, do not increase `Chunks` after every Queue. If the goal is six chunks, leave it at `6` until those six are complete.
#### 2. Generate and review the first chunk
1. Click the top-right blue `Run` button once.
2. Wait for the workflow, including video saving, to finish.
3. Play the resulting video. In the 5-second example, the first output is **5 seconds**, even though `Total Length` says 10 or 30 seconds.
4. Look at the Sampler. While more chunks remain, it should show `Chunk 1 is ready for review` and the actions below.
![Chunk 1 ready for review with continue, retry, finish, and Back to Settings buttons](images/v38-manual/live-review-actions-steps4.png)
The shorter output is intentional: Continuum is waiting for your decision before spending time on later chunks. If your total is only one chunk, it is already complete; there is no next chunk within that target.
`Render History — 68 Takes` in this screenshot is a count of stored Takes from that run's history. It does **not** mean the target is 68 chunks. Your count will differ.
#### 3. Keep the result and generate the next chunk
1. Click `Use it and continue` on the review panel.
2. Confirm the button now shows `✓ Use it and continue`. The card says `Selected: Use it and continue. Press Queue to run this action.`
![Use it and continue selected; Queue is still required](images/v38-manual/live-continue-selected-steps4.png)
3. Click the top-right blue `Run` button. Clicking the review button alone does not start generation.
4. Wait for completion and play the new output. The accepted chunks are reused; the next chunk is newly generated.
5. If the target is not yet complete, review the new chunk and repeat from step 1.
**Example:** With `Chunks = 6` and three 5-second chunks already completed, this action reuses Chunks 1–3 and generates Chunk 4. The saved video becomes 20 seconds, and you review Chunk 4 next. It does not immediately generate Chunks 5–6.
With the screenshot's `Chunks = 2`, continuing after Chunk 1 completes the target and produces a 10-second video. There is no third chunk unless you increase the target as described in [section 9](#9-add-more-chunks-to-a-completed-sequence).
![The two-chunk target is complete after only Chunk 2 is generated](images/v38-manual/live-sequence-complete-steps4.png)
At completion, `Try this chunk again` means “make another Take of the last reviewed chunk,” here Chunk 2. It does not mean that Continuum discarded Chunk 1. Use `Back to Settings` to change the target or `Render History` to inspect stored Takes.
#### 4. Try the current chunk again
Use this when the **chunk currently being reviewed** is the one you want to replace.
1. Keep `Base Seed` unchanged and `Control After Generate = fixed`. Do not switch to `randomize` to get another Take.
2. Click `Try this chunk again` on the review panel.
3. Confirm that this action has the `✓`, then click the top-right blue `Run` button.
4. Play the new result. Continuum keeps the earlier accepted chunks and creates another Take of the reviewed chunk, with its own variation managed internally.
5. If you like it, use `Use it and continue`. Otherwise, repeat `Try this chunk again` and the top-right blue `Run` button, or select an older Take from `Render History`.
**Example:** While reviewing Chunk 3 of a six-chunk target, retrying reuses Chunks 1–2 and generates a new Chunk 3. The output is still 15 seconds, not 20 or 30 seconds. Chunks 4–6 have not been generated on this path yet. The older Take remains in history.
![Try this chunk again selected; the top-right Run button executes the retry](images/v38-manual/live-retry-selected.png)
To replace an earlier chunk after moving past it, use [Render History](#7-choose-an-earlier-take-or-branch-from-it) or [Regenerate From](#8-regenerate-from-a-particular-chunk), not the current-chunk retry button.
If you change the seed, prompt, references or other generation inputs after a review, the panel shows **Settings changed** and withdraws the old Continue/Retry/Finish actions. Queue the current settings for the backend to decide compatible reuse, or select **Start again from Chunk 1** and then Queue for an explicit restart. This is non-destructive: existing Takes remain in Render History. If you manually select `Regenerate From = Chunk N`, press Queue directly; Review buttons no longer discard that boundary.
#### 5. Keep the result and finish everything remaining
1. Play and approve the chunk currently being reviewed.
2. Leave `Chunks` at the final total you want.
3. Click `Use it and finish the rest`.
4. Confirm its `✓`, then click the top-right blue `Run` button.
5. Wait for all remaining chunks and the final video to finish. This Queue does not pause for a review after each remaining chunk.
These three buttons are different even though all require `Queue`:
| While reviewing Chunk 3 of a six-chunk, 5-second-per-chunk target | Reused chunks | Newly generated chunks on the next Queue | Saved duration |
|---|---|---|---|
| `Use it and continue` | 1–3 | 4 | 20 seconds |
| `Try this chunk again` | 1–2 | A new Take of 3 | 15 seconds |
| `Use it and finish the rest` | 1–3 | 4–6 | 30 seconds |
For a two-chunk target after Chunk 1, `Use it and continue` and `Use it and finish the rest` produce the same remaining amount: only Chunk 2 is left. Their difference becomes clear with a larger target.
#### 6. Go back to settings without losing the review
1. On the review panel, click `Back to Settings`.
2. The normal controls appear again. You can inspect `Chunks`, `Run`, `Progress`, and the other settings.
3. Click `Show Advanced Settings` only if you need controls such as `Regenerate From` or `Run Name (Optional Override)`.
4. To go back without changing the operation, click `Return to Review`.
![Back to Settings keeps the completed sequence and exposes Return to Review](images/v38-manual/live-settings-complete-steps4.png)
![Return to Review above Regenerate From, Run Name, and Continuation Backend](images/v38-manual/return-to-review.png)
**Neither navigation button queues work, deletes a Take, resets the run, nor cancels a selected action.** `Return to Review` also does not undo settings you edited. Before the next Queue, select the action you actually want and check its `✓`.
`Show Advanced Settings` and `Hide Advanced Settings` only show or hide controls; they do not turn those settings on or off. `Back to Settings` is not a “start over” button.
#### 7. Choose an earlier Take or branch from it
A **Take** is a stored version of a generated result. A **branch** is an alternative continuation from a chosen result; the old history is retained.
1. On the review panel, click `Render History — … Takes`.
2. Use `Previous Take` and `Next Take` to choose a stored result. Read the displayed chunk and branch information; selecting an entry by itself does not replace your current sequence.
![Render History with Take navigation and the two explicit Take actions](images/v38-manual/live-render-history-steps4.png)
3. Choose one of the following actions, then click the top-right blue `Run` button:
- `Use This Take`: make the selected stored result the active result. This action runs no new sampling; the output path can still decode and save the selected sequence.
- `Continue From Here`: keep the sequence through the selected Take and generate the next review unit on a new continuation path.
4. Review the resulting output before continuing again.
**Example:** To keep an older Chunk 2 and make a new Chunk 3, select that Chunk 2 Take and use `Continue From Here`, then the top-right blue `Run` button. Do not select Chunk 3 if Chunk 3 is the part you want replaced. Later chunks from the old path are not automatically attached to the new one.
Keep `Regenerate From = Auto` when using Take actions. Do not combine an explicit `Regenerate From` selection with `Use This Take` or `Continue From Here`. If the selected Take already completes the configured total, `Continue From Here` has no next chunk to generate.
#### 8. Regenerate from a particular chunk
Use this when you know the **first chunk that must be redone**. Unlike retrying the current review, this can rebuild a longer section of an existing sequence.
1. Click `Back to Settings` if you are on the review panel.
2. Keep `Base Seed`, the total `Chunks`, and `Seconds per Chunk` unchanged. Keep `Control After Generate = fixed` and `Progress = On — Resume and Takes available`.
3. Click `Show Advanced Settings`.
4. Set `Regenerate From` to the first chunk to redo, such as `Chunk 3`. Earlier chunks can be reused only if their inputs remain compatible.
5. Set `Variation Nonce`, which appears for explicit regeneration. Keep `Base Seed` fixed. `0` selects a new variation automatically after a completed regeneration; a compatible interrupted regeneration resumes its existing variation instead. A value of `1` or higher fixes the variation: keep it to reproduce/resume that variation, or change it to request another. Repeating the same positive value with identical inputs does not request a different result. This is separate from `Try this chunk again`, which advances the review variation automatically.
![Regenerate From exposes Variation Nonce for an explicit regeneration](images/v38-manual/live-regenerate-from-chunk2.png)
6. Set `Run` to `Generate Full Video` to rebuild everything from the selected chunk through your target in one Queue.
![Run set to Generate Full Video while Progress stays On](images/v38-manual/full-video-progress.png)
7. Click the top-right blue `Run` button, wait for completion, and check the new full-length output.
8. Before ordinary continuation or review actions, set `Regenerate From` back to `Auto` so that the next Queue is not another explicit request to redo that section.
**Example:** With a completed six-chunk sequence, `Regenerate From = Chunk 3` and `Run = Generate Full Video` reuse compatible Chunks 1–2 and regenerate Chunks 3–6. This is **four newly generated chunks**, not only Chunk 3. Later chunks depend on the changed continuation and must follow the new path.
If you want a review pause instead, choose `Run = Review Each Chunk` in step 6. The next Queue generates the first requested review unit, not the whole remaining section. After reviewing it, select `Use it and continue` to proceed one unit at a time, or `Use it and finish the rest` to complete the section. Keep earlier-chunk inputs unchanged if you want those chunks reused.
#### 9. Add more chunks to a completed sequence
**You do not need Render History just to make a completed video longer.** For an ordinary extension, go back to the normal settings, increase the target `Chunks`, and Queue again. `Chunks` is the **final total number of chunks**, not the number you want to add next.
For the simplest and safest extension, keep the existing generation conditions unchanged: the same Sampler/workflow and run identity, `Base Seed`, `Control After Generate = fixed`, `Seconds per Chunk`, prompt, model/LoRA, media, dimensions, sampling schedule, Continuity, and Audio Continuity. Keep `Progress = On — Resume and Takes available` and `Regenerate From = Auto`.
**Example: extend a completed 2 × 5-second sequence to 3 × 5 seconds**
1. Finish the original two chunks. The completed target is `Chunks = 2`, so the saved video is 10 seconds.
2. Click `Back to Settings`. You do **not** need to open `Render History`.
3. Change `Chunks` from `2` to `3`. `Total Length` changes from 10 seconds to 15 seconds.
4. Keep `Progress = On — Resume and Takes available` and `Regenerate From = Auto`. Keep the prompt and the existing generation inputs unchanged for this ordinary extension.
5. Use `Run = Review Each Chunk` if you want to review the new chunk, or `Run = Generate Full Video` if you want all currently missing chunks generated in one Queue.
6. Click the top-right blue `Run` button.
With compatible saved progress, the expected work is:
- Chunk 1 → reused; no new Sampling
- Chunk 2 → reused; no new Sampling
- Chunk 3 → newly generated
- final saved sequence → 3 chunks / 15 seconds
The same rule applies to a larger extension. For example, changing `Chunks` from `3` to `5` means the new target is five total chunks; it does **not** mean “add five.” Review mode first generates Chunk 4, then Chunk 5 on the next accepted continuation. Full-video mode generates the missing Chunks 4–5 in one Queue.
Switching `Generate Full Video` ↔ `Review Each Chunk` by itself does **not** mean restart from Chunk 1. A Chunk 1 restart happens only when you explicitly choose `Start again from Chunk 1`, explicitly set Advanced `Regenerate From = Chunk 1`, or the backend determines that the stored prefix is incompatible with the current generation contract.
`Render History` is for a different job: browsing stored Takes, selecting an older Take, or branching with `Continue From Here`. Normal append-style extension does not require History. `Previous Take` / `Next Take` only browse stored results; they do not generate a new chunk.
**Do not assume append-only reuse with Last Image.** Extending a sequence changes where its final-image constraint belongs. The former last chunk, or a terminal pair, may need regeneration. Likewise, changing the prompt or other generation-contract inputs can invalidate saved progress. `Progress = On` enables compatibility checking; it does not force incompatible chunks to be reused.
The extension path is covered by CPU compatibility tests, including two-to-three chunks and Last Image invalidation. It is not the same GPU test as the two-chunk review timing example below; arbitrary extensions and media combinations have not all been GPU-verified.
#### 10. Understand continuation time and check reuse
Continuing does not mean saving only the new segment. With compatible saved progress, the normal workflow:
1. Reuses the accepted chunks' stored Video/Audio latents without sampling them again.
2. Samples the requested new chunk or remaining section.
3. Decodes the completed sequence so far, runs `H3 Continuum Finalize`, and saves the combined video again.
The full-output work grows as the sequence gets longer. Consequently, the next Queue can take as long as the first, or longer, even though earlier chunks are **not** being sampled again. Loading, preparation, decoding, assembly, saving, and other workflow overhead all contribute to total time.
The latest live GPU check used Turbo4, Euler/simple, actual `Steps = 4`, Spectrum Off, and `Draft — 0.30 MP` with `Chunks = 2`, `Seconds per Chunk = 5`, and Review enabled:
| Queue | Sampler work | H3 Continuum Sampler node time | Saved video |
|---|---|---:|---:|
| First | Generate Chunk 1 | 103.802 s | 5 seconds |
| After `Use it and continue` | Reuse Chunk 1; generate Chunk 2 | 43.738 s | 10 seconds |
Each Queue sampled one chunk with four steps. The second Queue did not sample Chunk 1 again; it sampled only the requested next chunk, then decoded, finalized, and saved the combined 10-second sequence. The first Queue also included cold-start/model preparation effects, so the difference must not be treated as a general speed ratio. Full-sequence decode and save work still occurs on continuation.
To inspect your own run:
1. Open ComfyUI Settings and search for `H3 Continuum: Detailed Report`.
2. Enable that setting before the Queue you want to inspect.
3. After completion, inspect the Sampler's `status` output through a text-display node in your workflow.
4. In the two-chunk example's second Queue, look for `1 reused, 1 generated, 2 total`. Detailed fields should show `sampled_physical_groups=1` and, with four steps, `sampling_steps_total=4`, not `8`.
The report field names are diagnostic text, not controls you need to set. For a larger target, use the reported reused/generated counts rather than timing alone. Terminal Merge can combine two chunks into one sampled group, so its group count differs from its chunk count.
#### 11. Troubleshoot a missing action or unexpected result
| What you see | What to check or do |
|---|---|
| A 5-second video despite a longer `Total Length` | In `Review Each Chunk`, this is the expected first output. Review it, select an action, then click the top-right blue `Run` button. |
| No next-chunk action | Check whether the configured total is already complete. Otherwise, wait for Queue completion and verify `Run = Review Each Chunk` and `Progress = On — Resume and Takes available`. If you opened settings, click `Return to Review`. |
| No review panel even though an incomplete Review run finished successfully | Inspect the Sampler's `status` and any ComfyUI error. This is not a reason to assume all chunks finished or to change `Base Seed`; the frontend must receive the saved review state. |
| Clicking an action appears to do nothing | Check its `✓`, then click ComfyUI's top-right blue `Run` button. The action button only selects what Queue will do. |
| `Set Control After Generate to fixed` | Set that exact control to `fixed` and retain the original `Base Seed` used for the saved chunks. |
| Settings disappeared behind the review panel | Click `Back to Settings`. Use `Return to Review` to come back; neither navigation button resets the run. |
| The next Queue takes as long as the first | Check reused/generated counts. Full-sequence decoding and saving still run; elapsed time alone cannot tell you whether chunks were resampled. |
| Earlier chunks are generated again | Check `Progress`, saved files, run identity, `Base Seed`, and changed generation inputs. Check that `Regenerate From` is `Auto` unless you intentionally requested regeneration. |
| Retrying changed more than one chunk | A terminal pair is one review unit. Also distinguish `Try this chunk again` from `Regenerate From`, which rebuilds the chosen chunk and its following section. |
| You want to stop choosing an action after every chunk | At the review panel, select `Use it and finish the rest`, then click the top-right blue `Run` button. Keep `Chunks` at your intended final total. |
Driving Audio can be used with Review and Smart Regenerate. The Continuum Image, Audio, and Video loaders support native bypass for optional media paths.
> **Second Pass limitation:** Finish the reviewed sequence before running Second Pass. Second Pass / `refine_context` is not supported on a partial Review sequence.
+309
View File
@@ -0,0 +1,309 @@
# H3 Continuum — 日本語の詳細ガイド
[日本語README](../README_JA.md) · [英語の図解ガイド](USER_GUIDE.md) · [日本語の履歴資料](README_HISTORY_JA.md)
この資料は、正本`84899c1`(パッケージ3.9.8)の日本語READMEから、詳しい操作と画像を移したものです。
V3.8の表示名・操作説明は保持しているV3.8X2用であり、V3.9の導入・Reference説明とは区別してください。
過去の版表記、修正記録、測定条件は当時の記録として保持しています。
現在の配布・互換性・制約は[短縮版README](../README_JA.md)と[KJ対応条件](KJ_PREVIEW_OVERRIDE_V398.md)を参照してください。
- [V3.9とV3.8X2の導入](#最初にv39とv38x2)
- [Reference Images](#v39-reference-images)
- [Review・継続の手順](#10秒を5秒ずつ確認しながら作る)
- [Prompt・スキル配布](#プロンプトスキルのダウンロード)
<!-- Preserved README sections begin. -->
## 最初に:V3.9とV3.8X2
ComfyUIの **Templates → ComfyUI-H3-Continuum** には、今回選んだ[V3.9公式Workflow](../examples/workflows/MiniMax_H3_Continuum_V39.json)と、別の[V3.8X2 Workflow](../examples/workflows/MiniMax_H3_Continuum_V38X2.json)を2件表示します。[V3.8X2 Workflow ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip)には、**V3.8X2とV3.9の両JSON**を同梱しました。[V3.9単独ZIP](../examples/workflows/MiniMax_H3_Continuum_V39.zip)も残します。いずれもWorkflow用ZIPで、カスタムノード本体のインストーラーではありません。V3.8 Samplerは残るので既存のV3.8X2 Workflowは使えます。ただしV3.9へReference配線や保存済みRun/Takeは自動移行しません。旧Workflowを保存しておき、新しい作業ではV3.9専用Workflowを開いてください。
選択したV3.9テンプレートは32ノードです。画像Loader 1~6は新ヘルパーへ配線済み・初期Bypassです。7~9番の入力欄は残し、必要な人が画像Loaderを追加して接続します。使用する画像を選んで必要なLoaderだけ有効にしてください。保存済みサイズ設定は**First Image+Draft — 0.30 MP**で、別のFirst Image Loaderは有効です。Queue前に自分のFirst Imageファイルを選び直してください。保存済みのWidth/Height 480×640はManual時だけ有効です。`Reference Image Size`は`Match Output`で、読み込み後のconditioningを縮小しますが元画像ファイル自体は縮小しません。チャンク別Referenceの効果を明瞭に試すなら、First ImageとLast ImageをOFFにし、Manualへ切り替えてReference Imagesだけを使うのが勧められます。First/Lastを入れるとそれ自体が動画を拘束し、Last Imageは最後の2つの論理チャンクを1つの生成groupに結合する場合があります。その場合、両チャンクの有効Reference集合は同じにしてください。Referenceを外しても、前チャンクの映像から人物が引き継がれることはあります。
ヘルパーの `All chunks` は接続画像を全チャンクに、`Per chunk` は画像×チャンクの表で指定します。このテンプレートは`Per chunk`で全チェックOFFから始まるため、Loaderを有効にした後、使用したいチャンクにもチェックしてください。空欄があっても番号はずれず、Image 1は常に`@R1`、Image 9は`@R9`です。Sequence Promptにはチャンクごとの動作を短く書き、そのチャンクで使う画像の`@R`タグを付けます。内部の`<Picture N>`番号はgroupごとに自動変換されます。使わないタグは警告しますが生成を止めません。まず短いチャンクと縮小済み画像で試してください。入力元画像はSamplerのReference Image Size設定より前にRAMを使うことがあります。
V3.9.3は`main`の現行ソースです。バージョンを固定したソースは[GitHub Releases](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/releases)から取得してください。ComfyUI Registryへの公開は別の作業です。過去のReleaseとtagも残します。設定済み1024×1024のReference-only GPU/APIテストは通っていますが、無改変の公式テンプレート初期設定とブラウザー保存・再読込は別の受入項目です。古い環境を厳密に再現するには対応する過去のRelease/tagとWorkflowを使用してください。現行mainにもV3.8 Samplerを残し、V3.8X2 Workflowを使えるようにしています。
`pyproject.toml`の宣言上はComfyUI `>=0.32.0`ですが、以前の実生成・GPU検証はComfyUI `0.34.2`、上記の限定したFixedキャッシュ比較は`0.38.0`で行いました。最低宣言版でV3.9の実機動作を確認済みという意味ではありません。以前のWindows CPU試験は`1696 passed / 1 skipped / 0 failed`、以前の別環境Linux再実行報告は`1534 passed / 3 skipped / 0 failed`で、新しいキャッシュ/外部INT試験は含みません。これらは無改変の公式テンプレートの実ブラウザー保存・再読込やGPU受入を代替しません。
## V3.9 Reference Images
### Reference Imagesノード:画像とチャンクを選ぶ
![H3 Continuum Reference Images V3.9ノードの固定スロットとチャンク別割り当て表](images/v39/v39-reference-images-node.png)
*この画像は操作例で、配布Workflowの初期割り当てではありません。* Image 1~9は固定スロットで、途中が空でもImage 1は常に`@R1`です。`Reference Use = Per chunk`にすると、使いたい画像とチャンクの交点をチェックできます。各行の`All`/`Off`は、その画像を全チャンクで選択/解除します。画像ではImage 1をChunk 1・2に割り当て、ほかの画像は別のチャンクに割り当てています。公式V3.9 Workflowは画像Loaderが初期Bypass、割り当てが全OFFです。実ファイルを選び、Loaderを有効化してから、使うチャンクにチェックしてください。
### Sampler V3.9:ヘルパーから1本で接続
![Reference Images Optional入力を1本持つH3 Continuum Sampler V3.9](images/v39/v39-sampler-node.png)
ヘルパーの緑色の`reference_images`出力を、Samplerの`Reference Images (Optional)`入力へ1本つなぎます。V3.8X2 Workflowは旧V3.8 Samplerと従来のReference接続を使用し、開くだけでV3.9配線へ変換されるわけではありません。この画像のサイズ設定は**Manual 480×640の操作例**で、公式V3.9 Workflowの保存値は**First Image+Draft 0.30 MP**です。`Reference Image Size = Match Output`は画像読込後のReference conditioningサイズを決め、元ファイルを事前縮小したり、動画の出力解像度を直接決めたりはしません。
V3.9では、固定番号の参照画像1~9を新しい **H3 Continuum Reference Images V3.9** ノードにまとめます。ノード内の表で「All chunks」またはチャンク別のチェックを選び、Samplerへは1本だけ接続します。`Sequence Prompt`で`@R1`と書けば固定の画像1を指します。有効なタグは各生成groupの実際の`<Picture N>`へ変換されます。両モードとも同じチャンク別処理を通ります。
Promptは実行を止めません。明示的なTimelineでも解析できなければ元の入力をFixedとして扱い、診断`H3C-P100`を出します。Timelineで指定のないチャンクは前のPromptを再利用し、`H3C-P101`で知らせます。どの画像を各チャンクに使うかはReference Imagesノードで選び、Prompt中の`@R1`~`@R9`は接続画像の固定番号に対応させてください。
同梱の`2 × 10秒`Workflowでは、時間見出しを必ず単独行にし、各チャンクの動作だけを短く書きます。対応する画像LoaderをONにしてチャンク欄へチェックを入れた場合の書式例です。
```text
[0-10s]
@R1 walks forward through a quiet hallway. The camera follows smoothly.
[10-20s]
@R2 enters from the right. The camera turns to follow @R2.
```
人物や動作は実際の画像に合わせて書き換えてください。これは書式例で、配布WorkflowのPromptは空欄です。`[0-10s]`と本文を同じ行には書かないでください。チェックされた画像は、Prompt中に`@R`を書かなくてもそのチャンクのconditioningに入ります。
[公式V3.9 Workflow](../examples/workflows/MiniMax_H3_Continuum_V39.json)は、今回指定された添付JSONを無改変で採用しました。[単独ZIP](../examples/workflows/MiniMax_H3_Continuum_V39.zip)も用意しています。V38X2のモデル・LoRA、Decode Cache、Finalize、Saveのフル構成を維持し、画像Loaderは1~6だけを配線済み・初期Bypassにしています。7~9番を使う場合は`H3 Continuum Load Image`を追加し、ヘルパーの同じ番号の入力へ接続してください。Sequence Promptの入力ノードと配線は残し、文章は空欄です。Queue前に自分のプロンプトを入力してください。保存設定は`Timeline`、2×10秒、`Full Run`(画面表示はGenerate Full Video)、First Image+Draft 0.30 MP、別Project IDです。従来の[V3.9 Reference Inline移行用Workflow](../examples/workflows/MiniMax_H3_Continuum_V39_Reference_Inline.json)は変更せず残します。チャンク別割り当て、First/Lastとの終端結合、保存・再読込、画像メモリの注意点は[操作ガイド](V39_REFERENCE_INLINE.md)を参照してください。Queue前にモデル・素材のファイル名を確認し、画像を有効化する場合は実ファイルを選択してください。7~9枚の参照は特に高負荷なので、最初は枚数と元画像サイズを抑えてください。新ノードを表示するにはComfyUI本体の完全再起動とブラウザーの強制再読込が必要です。無改変の公式テンプレートの実ブラウザー保存・再読込/GPU受入はまだ未実施です。GitHub Releaseの公開は、これらの受入やRegistry公開の完了を意味しません。
**互換性:** V3.8X2の既存Workflowは旧V3.8 Samplerと旧4~9ヘルパーで引き続き利用します。V3.9では新ヘルパーへの接続が必要です。旧Workflow上でSamplerの型だけ変えず、上記のV3.9専用Workflowを開いてください。以前のV3.9試作WorkflowにあったSampler直結1~3やSampler側の割り当て欄は配線と保存widgetの位置が異なります。V3.8の途中Run/Takeは自動移行しません。元Workflowを保存し、V3.9では別Run Nameを使ってください。
以前のV3.9操作系のブラウザー確認は、新しいヘルパーの保存・再読込確認とは別です。新ヘルパーは設定済みGPU/API実行の証拠がありますが、公式テンプレート初期設定の受入は別です。CPU/JS・Manifestも別に判定し、V3.9 GitHub Release/Registry公開はまだ行っていません。
**バージョン境界:** V3.8X2(package 3.8.3)とV3.9.0は別Workflowです。今回、新しいtag・GitHub Release・保守ブランチ・Registry版は作成せず、既存Releaseも削除しません。旧V3.8.0は既存の`v3.8.0` tagから取得できます。
## mainのLoader保存・復元修正(2026-09-21、3.8.3以降)
現行V3.8X2ワークフローは、音声を **Core Load Audio**、動画を **Core Load Video → H3 Continuum Video Adapter** に変更しました。Video Adapterは24fpsへの変換(Force Rate初期値24)とIMAGE/AUDIO出力だけを担当し、**Enableやファイル選択UIはありません**。動画を使わない場合はAdapter、または動画入力グループ全体をBypassしてください。Load VideoだけをBypassして必須入力のAdapterをONに残す構成は避けてください。
画像は **H3 Continuum Load ImageとEnable Imageを維持**します。Issue #23の継承されたmode setterへの委譲と、描画・保存中のwidget配列差し替え廃止を適用します。EnableはCoreの欄の後ろに配置し、画像ファイルとBypassの保存はCoreに委ねます。旧Audio/VideoノードIDは既存workflow用の非推奨互換ノードとして残します。V3.8X2の登録IDは互換用2個を含め10個で、V3.9では別IDのV3.9 SamplerとReference Imagesヘルパーを追加して計12個です。V3.8 Sampler設定・Sampling・Decode Cache・Finalize・既存Run Storage契約は変更しません。
更新後はComfyUI再起動と画面の再読み込みが必要です。既に保存JSONから消えたファイル名やOFF状態は推測で復元できないため、一度選び直して保存してください。既存ユーザーworkflowや保存済みTakeは自動変更しません。上流ノードの種類変更でRun Storageが新revisionになることはあるため、過去Runの再開には元workflowを保管してください。
検証範囲は[Loader修正記録](LOADER_PERSISTENCE_REPAIR.md)を参照してください。CPU回帰テストと実ブラウザ確認は別です。Windows Core 0.36.0 / frontend 1.53.6のタブ切替確認は未実施で、Release/tag・Registry公開も別作業です。以下の旧Loader画像は互換ノードの説明です。
## Timeline Video — main上のExperimental機能
従来の動画IMAGE入力を**`Timeline Video Frames`**と表示し、末尾に省略可能な設定**`Video Reference Mode`**を1項目追加しました。新しい設定を持たない既存Workflow/APIは、初期値の**`Repeat Reference`**で従来動作を維持します。
- **`Repeat Reference`**:従来のVideo Guide契約です。各physical生成groupで、同じ先頭側の上限付き参照を使います。短い素材を自動ループ/Stretchしません。
- **`Follow Timeline`**:各physical出力groupの時間位置に対応する24fps参照区間を使います。Continuationの過去prefixを二重消費しません。素材終了後は動画参照なしで続行し、利用可能な終端区間にはH3 frame-grid整列分だけ短いpaddingを行う場合があります。
これは`Prompt Format = Timeline`とは別の設定です。新しいsocketやLoaderは増やさず、元動画のAudioも自動使用しません。参照動画はsoft conditioningであり、motionをframe単位で完全コピーする機能ではありません。また、decode済みIMAGE batch全体をsystem RAMへ保持するため、長尺入力のRAM削減機能ではありません。
接続は既存のままです。
```text
Core Load Video → H3 Continuum Video Adapter → Timeline Video Frames
```
公式V3.8X2 Workflow本体は変更せず、比較用Experimental Workflowを別名で追加しています。
- [Follow Timeline JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Follow.json)/[ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Follow.zip)
- [Repeat Reference JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Repeat.json)/[ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2_Timeline_Experimental_Repeat.zip)
ローカル受入ではFull CPU `1424 passed / 1 skipped / 0 failed`、ブラウザ保存・再読込、Follow/Repeatの`2 × 5秒` GPUテストがPASSしました。両方とも704×416、24fps、240フレーム、映像10秒、32kHz stereo音声10秒で、OOM/NaN/allocation failure/crashはありません。FollowはChunk 2で後半区間へ切り替わり、Repeatは同じ先頭側参照を維持しました。RTX 5060 Ti 16 GBで観測した最大VRAMは約15.6~15.7 GiBで、16 GBの余裕は小さい条件です。これは機能・mode分離の受入結果であり、一般的な速度や主観画質の保証ではありません。詳細は[Timeline Video Experimental](TIMELINE_VIDEO_EXPERIMENTAL.md)を参照してください。
> **V3.8X2**はpackage `3.8.3`の製品名・Workflow名です。V3.8 Production Samplerを維持し、任意のReference Image 4~9と内蔵Decode Cache Helperを追加しています。旧V3.8.0は[`v3.8.0`タグ](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum/tree/v3.8.0)から導入できます。
## V3.8X2 公式Workflow
- **標準名:** [JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2.json)/[ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip)
- **Helper明示名:** [JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.json)/[ZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2+Decode_Cache_Helper.zip)
V3.8X2の2つのJSON名は**同じ公式graph**を収録した配布上の別名で、異なる設定のWorkflowではありません。どちらも内蔵Decode Cache Helperと9枚Reference Image用の接続を含み、Helperの保存値は`Auto / RAM 256MB / Disk 8GB / reset_token 0`です。標準名のV3.8X2 ZIPには別のV3.9 JSONも同梱し、Helper明示名ZIPは対応するV3.8X2 JSONだけを収録します。既存projectでは従来のV3.8X Workflowも引き続き使用できます。
## Reference Image 9枚使用時のメモリ注意
Reference Image 1~3はSamplerへ直接接続し、任意の4~9は`H3 Continuum Reference Images`を経由して`Reference Images (Optional)`へ接続します。9枚すべてを使う構成は高負荷であり、通常の最低要件ではありません。
受入テストでは、**RTX 5060 Ti 16 GB/RAM 64 GB**環境で、9枚の元画像を各**約0.30 MP**に抑えた条件で完走しました。`512 × 608`出力、`3 × 5秒`、Spectrum Off、LoRA Off、Sage Attention、Balanced 22-frame continuityで、Sampling中のVRAMは`16,311 MiB`中の約**15.0~15.5 GiB**でした。16 GB環境では余裕がほとんどありません。
多数、特に9枚を使う場合は、読み込み・接続前に元画像のコピーを約0.30 MPへ縮小してください。Sampler内の通常のconditioning用Resizeは引き続き行われますが、小さい入力ファイルにすることで読み込み・前処理時の負荷を抑えられます。高解像度参照、大きな出力、追加wrapper、同時GPU処理では、より高性能なGPUと多いシステムRAMが必要になる可能性があります。9枚は任意であり、3枚より常に高画質になる保証はありません。
## 内蔵 Decode Cache Helper
### 重要:新規生成が一律に速くなる機能ではありません
**キャッシュの手動クリア**:普段は操作不要です。Helperの「キャッシュをクリア」ボタンを押すと、次回QueueでこのHelperのキャッシュを破棄し、通常Decodeからやり直します。ボタンは即時削除や自動Queueを行いません。内部の`reset_token`は保存・API互換用に残り、押すたびに1増えます(INT上限では0へ戻ります)。作成・読込・複製だけでは変更しません。JavaScriptが使えない環境やAPI版では、従来どおり`reset_token`の値を変更してください。
**短縮するのは、変更していないlatentを再びデコードする時間です。Sampling(生成計算)自体は高速化しません。** 初回は通常Decodeとキャッシュ保存を行うためMISSとなり、保存・照合の負荷で遅くなる場合があります。毎回ランダムシードで異なるlatentを新規生成する使い方では、基本的にMISSとなり、この機能による短縮は期待できません。
具体的に再利用できる場面:
- **Continueでチャンクを追加**:既存チャンクのlatentを保持し、出力時にもう一度Decodeする場合、保存済みの既存部分だけを再利用します。追加チャンクは初回MISSです。
- **Fix/部分再生成**:変更していないチャンクのlatentがそのまま再入力される場合、その部分だけを再利用します。再生成してlatentが変わった部分はMISSです。
- **Complete/Resumeで再出力**:Samplingをやり直さず、保存済みlatentから全体を再Decodeする場合に有効です。単なる既存動画ファイルの再生ではありません。
- **Decode以降のリトライ、Finalizeや出力設定の変更**:Helperへ同じlatent・Decode関連情報が再入力され、Decodeが再実行される場合に有効です。
- **固定シードで同条件を再生成**:生成後のlatentが完全に一致した場合だけ再利用できます。同じシードだけではHITを保証しません。
これらの操作名そのものが有効条件ではありません。**各Video/Audioの入力latentの内容・shape・dtype、Decode関連メタデータ、VAE識別情報が一致し、そのHelperにキャッシュが残っていること**がHITの条件です。ComfyUIが上流/出力をそのまま再利用してDecode自体を実行しない場合、Helperによる追加のDecode短縮はありません。Helperを接続し、`Auto`または`RAM`で使ってください。レポートの`hit`は再利用、`miss`は通常Decode、`off`/`bypass`はキャッシュ再利用なしを表します。
キャッシュはPythonプロセス内の利用に限られ、再起動後は再利用しません。上限による退避・破棄やVAE識別情報の変化でもMISSになります。モード、RAM/Disk予算、`reset_token`を変更すると、この実装ではHelperキャッシュをクリアします。`reset_token = 0`を維持することは、毎回リセットする意味ではありません。
`H3DecodeCacheHelper`をContinuumパッケージ内に収録し、UI上は`MiniMax H3/Continuum/Helpers`の独立公開ノードとして残します。インストール対象はContinuumだけです。V3.8X2の公開IDは、旧互換Loader 2件を含め計10件です。V3.9は別IDのSamplerとReference Imagesヘルパーを追加して計12件です。V3.8 Sampler、Finalize、Assembly Plan、Core Decode、保存済みwidget/socket契約は変更しません。
Samplerの`video_latents`/`audio_latents`と同じVideo/Audio VAEをHelperへ接続し、Helperの`images`/`audio`をFinalizeへ接続します。Samplerの`assembly_plan`はFinalizeへ直接接続したままです。[公式V3.8X2 Workflow](../examples/workflows/MiniMax_H3_Continuum_V38X2.json)は保存済みgraphと設定を維持します。このgraphにはCore VAE Decodeノードがないため、Core直結へ戻すにはCore Video/Audio Decodeを追加してFinalizeへ再配線するか、旧V3.8XのCore直結graphを開いてください。Helperを単に削除しても自動では戻りません。
`Auto`は全physical groupのVideo Decodeをプロセス専用Diskへ、小さなAudio Decodeを上限付きRAMへ保存します。`Off`はnative Decodeを使い、Helper Cacheを破棄します。Cache障害時はnative Decodeへ戻りますが、native DecodeのOOMやQueue中断は再試行しません。Python再起動後にCacheファイルを再利用しません。移行時は同じ`H3DecodeCacheHelper` IDを提供する旧単独addonと内蔵版を同時登録しないでください。内蔵版の導入確認後に旧addonを無効化または撤去します。
736×416(0.306MP)、3×5秒、Euler/simple 6 steps、Turbo FL2V v1.2、ウォーム、Video 3/3 HITの1条件では、Full Run Totalの**機構推定中心値は約47秒/約24.5%短縮**でした。直接A/Bの1ペアは50.887秒/26.46%、Samplingを伴わないComplete/Resumeの2ペアは平均46.309秒/79.826%短縮でした。これは限定条件の観測値で、反復された一般平均や無条件の20%以上高速化保証ではありません。初回MISSは遅くなる場合があり、新規Samplingは高速化せず、latentまたはVAE識別情報が変わればMISSになります。
MiniMax H3を複数チャンクで連続生成し、直前チャンク末尾の**映像latent / 音声latentを直接**次チャンクへ継承するComfyUIカスタムノードです。チャンク間でVideo/Audio VAEのDecode→Encodeは行いません。
## インストール・更新
新規導入:
```bash
cd ComfyUI/custom_nodes
git clone https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum.git
```
既存のGit導入を更新:
```bash
cd ComfyUI/custom_nodes/ComfyUI-H3-Continuum
git pull --ff-only origin main
```
ComfyUIを再起動してください。Managerから導入した場合はManagerのUpdateを使い、同じノードを重複配置しないでください。
旧V3.8.0を別フォルダへ導入する場合:
```bash
cd ComfyUI/custom_nodes
git clone --branch v3.8.0 --single-branch https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum.git ComfyUI-H3-Continuum-v3.8.0
```
Continuumのcheckoutは同時に1つだけ有効にしてください。ComfyUI Managerの**Update**は`main`を更新するため、過去tagの選択には使いません。
V3.8X2 Workflow:[JSON](../examples/workflows/MiniMax_H3_Continuum_V38X2.json)/[V3.8X2・V3.9両JSONを含むZIP](../examples/workflows/MiniMax_H3_Continuum_V38X2.zip)。V3.8X2 JSONはSpectrum対応の1本で、保存状態ではSpectrumとTurbo LoRAはどちらもOFFです。[LightX2V Turbo](https://github.com/ModelTC/Minimax-H3-Turbo)にも切り替えられます。完全なgraphを開くにはSpectrum・rgthree・KJNodesが必要です。ComfyUI-Easy-Useは不要です。
## プロンプト・スキルのダウンロード
- [V3.9 Reference Images用プロンプトスキル本文](../write-h3-v39-reference-prompts/SKILL.md)/[ZIP](../write-h3-v39-reference-prompts.zip):固定の`@R1`~`@R9`とチャンク別の画像割り当てに対応します。V3.9のReference Imagesノード用で、下記の汎用Continuumスキルとは別です。
- [LLM用システムプロンプトZIP](../H3-Continuum-LLM-System-Prompt-v1.zip):Continuum向けプロンプト作成のシステム指示と詳細資料。
- [Continuum専用プロンプトスキルZIP](../H3-Continuum-Skill-v1.zip):Codexなどで使う、チャンク構成に対応した汎用プロンプト作成スキル。
- [Continuum Dance DirectorスキルZIP](../minimax-h3-continuum-dance-director.zip):主役のダンサー1人を対象に、チャンク間の身体動作とカメラの連続性を考慮した長尺ダンス用プロンプトを作成するスキル。
いずれも任意のプロンプト作成支援資料で、ComfyUIのカスタムノードではありません。ZIPを展開し、同梱の案内に従って利用してください。Samplerを変更するものではなく、生成動作や画質を保証するものでもありません。
## V3.8の公開範囲
H3 Continuumは、長尺映像を最初からやり直さずに生成・確認・部分再生成・再開するProduction Samplerです。V3.8は`Main / Production`と`Advanced`の2層に整理しました。
以下の9ノードに`H3 Continuum Video Adapter`を加えた計10 IDがV3.8X2の登録面です。旧Audio/Video Loaderは互換IDとして残ります。V3.9はさらにSampler V3.9とReference Images V3.9を追加した計12 IDです。
- `H3 Continuum Sampler V3.8`
- `H3 Continuum Finalize`
- `H3 Continuum Load Image`
- `H3 Continuum Load Audio`
- `H3 Continuum Load Video`
- `H3 Continuum Second Pass`
- `H3 Continuum Reference Audios`
- `H3 Continuum Reference Images`
- `Decode Cache Helper`
Main Sampler内の`Show Advanced Settings`/`Hide Advanced Settings`で詳細設定を開閉します。旧node property `H3 Continuum View`を含む保存Workflowは対応する開閉状態へ移行し、旧propertyを削除します。これは表示状態だけの変更です。Python input、widget順、保存済みwidget値、backend引数、Sampling identity、Run Storage identityは変更しません。Frontend extensionが読み込まれない場合も、Python定義の全widgetが表示された状態で実行できます。
V3.8X2公式Workflowの標準経路は`Sampler V3.8 -> Decode Cache Helper -> Finalize -> Create Video -> Save Video`です。HelperはMISSしたentryをCore Video/Audio Decodeへ委譲します。Core Decodeへ直接つなぐ手動構成も使用できますが、Decode結果のcacheはありません。Finalizeの`IMAGE`/`AUDIO`出力をCoreの`Create Video`で`VIDEO`へまとめてから保存します。Save、Upscalerは外付けとし、外部latent processorやupscalerはSecond Passをbridgeとして接続します。外部processorが変更できるのはVideo LATENTの空間解像度だけで、physical group、B/C/T、finite値、First Pass Audio、元のAssembly Planを維持する必要があります。Finalizeは公開`IMAGE`/`AUDIO`型を受け取り、特定decoder classへ依存しません。SageAttention、Sol-Attn、Spectrumは外部MODEL wrapperのままです。詳細は[V3.8 Open Integration Contract](V38_OPEN_INTEGRATION_CONTRACT.md)を参照してください。旧Hi-Res FixはV3.8X2標準Workflowに含めません。
### Reference Audio 1/2/3
`H3 Continuum Reference Audios`は、最大3本の単独Reference Audioを1本の`Audio References (Optional)` socketへまとめます。入力は間を空けずに接続し、Promptでは接続順どおりに`<Audio 1>`、`<Audio 2>`、`<Audio 3>`を使用します。共通のCore Audio VAEで各Audioを個別encodeします。生成後の最終Audioを置換せず、Driving Audio契約も変更しません。既存Workflowは従来の単数`Reference Audio (Optional)`をそのまま使用できますが、単数経路とbundle経路は同時接続しないでください。
> **対応範囲:** 元のV3.8公開面は7 ID、V3.8X2は互換Loaderを含め10 ID、V3.9は12 IDです。Finalizeの`H3ContinuumAssembleSeamV35`、Second Passの`H3ContinuumSecondPassV35`などは旧IDを維持しますが、すべての旧Workflowに互換性があるという意味ではありません。未対応のnode IDには対応するhistorical Release/tagを使用してください。詳細は[V3.8 Release/Migration Policy](V38_RELEASE_AND_MIGRATION.md)を参照してください。
複雑な16GB GPU受入Gateでは約`15.5~15.6 GiB`を使用しました。GPU、driver、backend、model精度、解像度、接続ノードで変動するため、すべての16GB GPUでの動作保証ではありません。
Reference Imageは最大9枚です。Samplerの直入力1~3を維持し、追加4~9は`H3 Continuum Reference Images`のoptional IMAGE入力へ接続して、出力をSamplerの`Reference Images (Optional)`へ渡します。補助ノードは画像を束ねるだけで、Resize・hash・VAE Encodeは既存Sampler経路で実行します。未接続を除き1→9の順に使い、Picture番号はFirst/Last Imageの後に連番で割り当てます。空のbundleは参照なしとなり、旧0~3枚の生成・再利用契約は変わりません。
以前の4・5直入力を含む保存Workflowは、グラフ読み込み後に元の画像接続をbundleへ移します。新接続を検証した後だけ旧接続口を削除し、接続元が不明・移行先が使用中の場合は警告して元の線を残します。旧API入力も維持しますが、同じ番号の画像を旧直入力とbundleの両方から指定すると入力競合として明示します。既存の3枚loaderテンプレートは上書きせず、そのまま利用できます。
### サイズの決め方を直接選択
| Preset | 目安 |
|---|---|
| `Draft` | 約0.30 MP |
| `Balanced` | 約0.60 MP |
| `Native 768` | 短辺768を目標。極端なAspectでは長辺1344 capを優先 |
| `Custom MP` | Megapixelを直接指定。Native 768のcapとは別 |
Mainの`Size Source`は次の2択です。
- `First Image`:接続したFirst Imageの縦横比を維持し、Presetの目標面積から最終Width/Heightを算出します。
- `Manual`:Width/Heightを32 pixel単位で直接指定します。この場合、Presetは適用しません。
MainのWidth/Heightは`Manual`のときだけ表示・編集できます。`First Image`ではResolutionを表示し、画像の縦横比からサイズを決めます。有効なFirst ImageがSamplerへ届かない場合は、保持しているManual Width/Heightへフォールバックし、その寸法をStatusへ表示します。手動寸法を確認・変更する場合は`Size Source = Manual`を選びます。Reference ImageやVideo Guideからサイズを暗黙取得することはありません。
旧V3.8の`Auto / Landscape / Portrait / Square`はWorkflow/API互換用として内部で受け付けます。FrontendはAutoをFirst Imageへ、3つの固定Aspectを同じ解決結果のManual Width/Heightへ移行します。
### 10秒を5秒ずつ確認しながら作る
ComfyUI上端の実行ボタンは、frontendにより`Queue`、`Run`、または`実行する`と表示されます。この手順内の`Queue`は、その実行ボタンを指します。
ここでは、画面に表示される名前だけを使って手順を説明します。
#### 最初に知っておくこと
`Chunks = 2`、`Seconds per Chunk = 5`は、完成目標が合計10秒という意味です。ただし、`Run = Review Each Chunk`では10秒を一度に生成しません。最初のQueueでは、確認用に**Chunk 1の5秒だけ**を生成します。これは正常動作です。
Chunk 1を確認した後に操作を選び、もう一度QueueするとChunk 2へ進みます。`Chunks = 1`では最初の5秒が最後なので、次へ進む操作はありません。
#### 手順1:生成前の設定
`H3 Continuum Sampler V3.8`で、上から次の順に設定します。
1. `Control After Generate`を`fixed`にします。`randomize`、`increment`、`decrement`では続きを再利用できません。
2. `Chunks`を`2`にします。
3. `Seconds per Chunk`を`5`にします。
4. `Total Length`が`10 seconds`になったことを確認します。`Total Length`は確認表示で、直接変更する項目ではありません。
5. `Run`を`Review Each Chunk`にします。
6. `Progress`を`On — Resume and Takes available`にします。
7. 緑色の`Ready to Queue`欄に、`2 × 5s = 10 seconds`と`Review each chunk`が表示されていることを確認します。`Set Control After Generate to fixed`と表示された場合は、Queueする前に手順1を直します。
#### 手順2:最初の5秒を作る
1. ComfyUIの`Queue`を押します。
2. Chunk 1だけが生成されます。出力動画が5秒でも正常です。
3. Queueの処理が完了すると、Samplerの表示が自動的に`Chunk 1 is ready for review`へ切り替わります。
#### 手順3:Chunk 1の動画を確認する
保存された5秒の動画を再生し、そのまま採用するか、作り直すかを決めます。この時点ではまだ合計10秒の動画は完成していません。
#### 手順4:次の操作を1つ選ぶ
| 画面上のボタン | 次のQueueで行うこと |
|---|---|
| `Use it and continue` | 現在の5秒を採用し、次のChunkを1つ生成します。 |
| `Try this chunk again` | 現在のChunkだけを別のTakeとして作り直します。 |
| `Use it and finish the rest` | 現在までを採用し、残りのChunkをまとめて生成します。 |
ボタンを選んだだけでは生成は始まりません。選択したボタンに`✓`が付いたことを確認してから、ComfyUIの`Queue`を押します。
`Try this chunk again`はContinuum側で別Takeを作るため、作り直しでも`Control After Generate`は`fixed`のままにしてください。
#### 手順5:2本目の5秒を作り、10秒を完成させる
1. `Use it and continue`を選びます。
2. ComfyUIの`Queue`をもう一度押します。
3. 採用済みのChunk 1を再利用し、Chunk 2を生成します。
4. Samplerから後ろのDecode、`H3 Continuum Finalize`、Save Videoまで通常どおり実行されると、合計10秒の完成動画が出力されます。
設定を確認し直したい場合は`Back to Settings`を押します。未処理のReviewへ戻る場合は`Return to Review`を押します。どちらも生成を自動開始せず、選択済みのTakeや保存済みChunkも削除しません。
#### 迷ったときの確認
- 最初のQueueで5秒だけ出た:`Review Each Chunk`の正常動作です。
- 次へ進むボタンが出ない:Queueが完了しており、`Progress = On — Resume and Takes available`、`Chunks = 2`以上であることを確認してください。正常なら完了後に自動表示されます。
- ボタンを押しても生成されない:操作を選んだ後、ComfyUIの`Queue`をもう一度押してください。
- `Set Control After Generate to fixed`と表示された:`Control After Generate`を`fixed`にしてください。別SeedのChunk 1が増えることを防ぐ安全チェックです。
- 1回で10秒すべて作りたい:`Run`を`Generate Full Video`にします。
Long Terminal Mergeのpairは分割せず、1つのReview単位として扱います。
Driving AudioはReview/Smart Regenerateと併用できます。Continuum Image/Audio/Video Loaderは任意入力向けのnative bypassに対応します。
> **V3.8 Reviewの制限:** Review途中のpartial sequenceでは、Second Pass/`refine_context`を正式対応範囲に含めません。Reviewを完了してsequenceを確定してからSecond Passを実行してください。V3.9では、上記のとおり検証済みの出力区間条件を継承できます。
### Take・Branch・安全な継続
Run Storageを有効にすると、Production表示に**Render History / Takes**が表示されます。
Takeを選択してもcanonical結果は変わりません。**Use This Take**でそのrevisionを
canonicalにし、**Continue From Here**で選択したTakeまでを保持して後続groupだけを
再生成します。どちらの操作も自動Queueは行いません。
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
+2 -2
View File
@@ -1,4 +1,4 @@
[project]
name = H3 Continuum
version = 3.9.4
description = Native MiniMax H3 masked AV continuation, Second Pass, and disk-backed assembly.
version = 3.9.10
description = Native MiniMax H3 masked AV continuation, Second Pass, and disk-backed assembly.
+1 -22
View File
@@ -440,14 +440,6 @@ from .v3.latent_resize_nodes import (
NODE_CLASS_MAPPINGS.update(V35_RESIZE_NODE_CLASS_MAPPINGS)
NODE_DISPLAY_NAME_MAPPINGS.update(V35_RESIZE_NODE_DISPLAY_NAME_MAPPINGS)
from .v3.tiled_latent_upscale_nodes import (
NODE_CLASS_MAPPINGS as TILED_UPSCALE_NODE_CLASS_MAPPINGS,
NODE_DISPLAY_NAME_MAPPINGS as TILED_UPSCALE_NODE_DISPLAY_NAME_MAPPINGS,
)
NODE_CLASS_MAPPINGS.update(TILED_UPSCALE_NODE_CLASS_MAPPINGS)
NODE_DISPLAY_NAME_MAPPINGS.update(TILED_UPSCALE_NODE_DISPLAY_NAME_MAPPINGS)
from .v3.hires_fix_nodes import (
NODE_CLASS_MAPPINGS as V35_HIRES_FIX_NODE_CLASS_MAPPINGS,
NODE_DISPLAY_NAME_MAPPINGS as V35_HIRES_FIX_NODE_DISPLAY_NAME_MAPPINGS,
@@ -461,15 +453,9 @@ from .v3.video_adapter_nodes import H3ContinuumVideoAdapter
NODE_CLASS_MAPPINGS["H3ContinuumVideoAdapter"] = H3ContinuumVideoAdapter
from .decode_cache.nodes import H3DecodeCacheHelper
from .prompt_builder import (
NODE_CLASS_MAPPINGS as PROMPT_BUILDER_NODE_CLASS_MAPPINGS,
NODE_DISPLAY_NAME_MAPPINGS as PROMPT_BUILDER_NODE_DISPLAY_NAME_MAPPINGS,
)
NODE_CLASS_MAPPINGS["H3DecodeCacheHelper"] = H3DecodeCacheHelper
NODE_DISPLAY_NAME_MAPPINGS["H3DecodeCacheHelper"] = "Decode Cache Helper"
NODE_CLASS_MAPPINGS.update(PROMPT_BUILDER_NODE_CLASS_MAPPINGS)
NODE_DISPLAY_NAME_MAPPINGS.update(PROMPT_BUILDER_NODE_DISPLAY_NAME_MAPPINGS)
# V3.8 exports only its audited launch surface. The imports and merged mappings
# above intentionally stay in place because the V3.8 implementation still
@@ -487,16 +473,9 @@ _PUBLIC_NODE_DISPLAY_NAMES = {
"H3ContinuumVideoAdapter": "H3 Continuum Video Adapter",
"H3ContinuumSecondPassV35": "H3 Continuum Second Pass",
"H3DecodeCacheHelper": "Decode Cache Helper",
"H3ContinuumTiledLatentUpscalePerClip": (
"H3 Continuum Tiled Latent Upscale (Per Clip)"
),
"DumasH3SequencePromptBuilder": "H3 Sequence Prompt Builder",
}
NODE_CLASS_MAPPINGS = {
node_id: NODE_CLASS_MAPPINGS[node_id]
for node_id in _PUBLIC_NODE_DISPLAY_NAMES
}
NODE_DISPLAY_NAME_MAPPINGS = {
node_id: f"DUMAS {display_name}"
for node_id, display_name in _PUBLIC_NODE_DISPLAY_NAMES.items()
}
NODE_DISPLAY_NAME_MAPPINGS = dict(_PUBLIC_NODE_DISPLAY_NAMES)
+19
View File
@@ -0,0 +1,19 @@
{
"id": "kj-preview-upstream-17195fc-phase-a",
"source_sha256": "1916295463eb790d3e8ea2b0dae2b25344549479b92e90a1d238acddb2dde3ab",
"siblings": {
"tiny_vae": {
"load_tiny_vae_decoder": "load_tiny_vae_decoder"
},
"ltxv_nodes": {
"_LTXWrappedPreviewer": "WrappedPreviewer",
"_ltx_rgb_factors": "get_ltx_rgb_factors"
}
},
"sibling_sha256": {
"tiny_vae": "61bf4daa634ef6bb901ff8630071caed12e733d17e7c4ba50082dc0aa5506591",
"ltxv_nodes": "11286583c56e653dbd57f5e8eda6902c84b759f16b359b3bf0c895b43e6b5364"
},
"core_preview_sha256": "b323be881ade3388254a72f5463e329f6d0a263b0078b69ae4256eb4e22dbba2",
"core_dispatch_sha256": "1c7b3b855278d0b95d1c1a705fa0b3c91a265f320e2f97bdeb9d15469d6d4ac7"
}
+19
View File
@@ -0,0 +1,19 @@
{
"id": "kj-preview-1.5.2-local-d07c8e5-phase-a",
"source_sha256": "d07c8e5ff7d7abd3e31acefe2cf29f618b5da5341b3b434e72ec2ac0953230b0",
"siblings": {
"tiny_vae": {
"load_tiny_vae_decoder": "load_tiny_vae_decoder"
},
"ltxv_nodes": {
"_LTXWrappedPreviewer": "WrappedPreviewer",
"_ltx_rgb_factors": "get_ltx_rgb_factors"
}
},
"sibling_sha256": {
"tiny_vae": "61bf4daa634ef6bb901ff8630071caed12e733d17e7c4ba50082dc0aa5506591",
"ltxv_nodes": "11286583c56e653dbd57f5e8eda6902c84b759f16b359b3bf0c895b43e6b5364"
},
"core_preview_sha256": "b323be881ade3388254a72f5463e329f6d0a263b0078b69ae4256eb4e22dbba2",
"core_dispatch_sha256": "1c7b3b855278d0b95d1c1a705fa0b3c91a265f320e2f97bdeb9d15469d6d4ac7"
}
+294
View File
@@ -0,0 +1,294 @@
"""Bounded KJ Preview Override identity; no provider imports or preview execution.
Only the source profiles shipped with this adapter can omit ``frames``. Other
wrappers keep the ordinary observer, including its existing safety decisions.
"""
from __future__ import annotations
import ast
import hashlib
import json
import marshal
import sys
import types
from functools import lru_cache
from pathlib import Path
_PROFILE = json.loads(Path(__file__).with_name("preview_override_provider_v1.json").read_text(encoding="utf-8"))
def _load_public_profile(path):
# A partial update must not break the old profile or ordinary generation.
try:
profile = json.loads(Path(path).read_text(encoding="utf-8"))
if (type(profile) is not dict or profile.keys() != _PROFILE.keys()
or type(profile["id"]) is not str or type(profile["source_sha256"]) is not str
or profile["id"] == _PROFILE["id"]
or profile["source_sha256"] == _PROFILE["source_sha256"]):
return None
return profile
except (OSError, ValueError):
return None
_PUBLIC_PROFILE = _load_public_profile(Path(__file__).with_name("preview_override_provider_public_v1.json"))
_PROFILES = (_PROFILE,) + ((_PUBLIC_PROFILE,) if _PUBLIC_PROFILE is not None else ())
_FIELDS = frozenset(("max_resolution", "node_id", "jpeg_quality", "suppress_default",
"preview_frames", "preview_fps", "vae", "tiny_vae", "audio_vae", "frames"))
_PREVIEW_REASONS = {
"provider source is outside the audited profile": "KJ/Core preview implementation is outside the audited source profiles",
"VAE preview is outside Phase A": "KJ preview VAE/Audio VAE or Tiny VAE is outside the supported configuration",
"Core approximate VAE is outside Phase A": "Core approximate-VAE preview is outside the supported configuration",
}
_UNKNOWN_PREVIEW_REASON = "KJ preview runtime code or state could not be verified"
PREVIEW_REUSE_REASONS = frozenset((*_PREVIEW_REASONS.values(), _UNKNOWN_PREVIEW_REASON))
def _normalized_code(code):
return code.replace(co_filename="<audited-provider>", co_consts=tuple(
_normalized_code(value) if isinstance(value, types.CodeType) else value
for value in code.co_consts))
def _literal(node):
# The audited source has one literal default expressed as 256 << 20.
if isinstance(node, ast.BinOp) and isinstance(node.op, ast.LShift):
return ast.literal_eval(node.left) << ast.literal_eval(node.right)
return ast.literal_eval(node)
@lru_cache(maxsize=8)
def _compiled(source):
tree = ast.parse(source)
code = compile(source, "<audited-provider>", "exec", dont_inherit=True)
codes = {}
def collect(item):
for child in item.co_consts:
if isinstance(child, types.CodeType):
codes[child.co_qualname] = child
collect(child)
collect(code)
functions = {}
classes = {}
for node in tree.body:
if isinstance(node, ast.FunctionDef):
functions[node.name] = node
elif isinstance(node, ast.ClassDef):
classes[node.name] = node
return codes, functions, classes
def _source(path, expected):
data = Path(path).read_bytes()
# Match the audited file contents across Git's LF/CRLF checkout policies.
data = data.replace(b"\r\n", b"\n")
if hashlib.sha256(data).hexdigest() != expected:
raise ValueError("provider source is outside the audited profile")
return data.decode("utf-8-sig")
def _provider_source(path):
data = Path(path).read_bytes().replace(b"\r\n", b"\n")
digest = hashlib.sha256(data).hexdigest()
matches = [profile for profile in _PROFILES if profile.get("source_sha256") == digest]
if len(matches) != 1:
raise ValueError("provider source is outside the audited profile")
return matches[0], data.decode("utf-8-sig")
def _check_function(fn, node, code, namespace):
if (type(fn) is not types.FunctionType or fn.__globals__ is not namespace
or fn.__closure__ or vars(fn) or fn.__qualname__ != code.co_qualname
or fn.__module__ != namespace["__name__"]):
raise ValueError("unknown loaded callable")
if marshal.dumps(_normalized_code(fn.__code__)) != marshal.dumps(_normalized_code(code)):
raise ValueError("loaded code differs from audited source")
defaults = tuple(_literal(value) for value in node.args.defaults) or None
kwdefaults = {arg.arg: _literal(value) for arg, value in zip(node.args.kwonlyargs, node.args.kw_defaults) if value is not None} or None
if not _same_literal(fn.__defaults__, defaults) or not _same_literal(fn.__kwdefaults__, kwdefaults):
raise ValueError("loaded defaults differ from audited source")
def _same_literal(actual, expected):
if type(actual) is not type(expected):
return False
if type(expected) in (tuple, list):
return len(actual) == len(expected) and all(_same_literal(a, b) for a, b in zip(actual, expected))
if type(expected) is dict:
return actual.keys() == expected.keys() and all(_same_literal(actual[key], value) for key, value in expected.items())
return type(expected) in (type(None), bool, int, float, str) and actual == expected
def _check_plain_class(cls, node, codes, namespace, *, sentinel=False):
if (type(cls) is not type or cls.__bases__ != (object,)
or cls.__module__ != namespace["__name__"]):
raise ValueError("unknown wrapper class")
methods = {item.name: item for item in node.body if isinstance(item, ast.FunctionDef)}
allowed = set(methods) | {"__module__", "__doc__", "__dict__", "__weakref__",
"__firstlineno__", "__static_attributes__"}
if sentinel:
allowed.add("_STOP")
if type(vars(cls).get("_STOP")) is not object:
raise ValueError("unknown encoder sentinel")
if set(vars(cls)) - allowed:
raise ValueError("additional class attributes")
for name, method in methods.items():
_check_function(vars(cls).get(name), method, codes[cls.__name__ + "." + name], namespace)
def _check_selected(module, source, names):
codes, functions, classes = _compiled(source)
for name in names:
value = vars(module).get(name)
if name in functions:
_check_function(value, functions[name], codes[name], vars(module))
continue
node = classes[name]
if not isinstance(value, type) or value.__module__ != module.__name__:
raise ValueError("unknown imported class")
methods = {item.name: item for item in node.body if isinstance(item, ast.FunctionDef)}
allowed = set(methods) | {"__module__", "__doc__", "__dict__", "__weakref__",
"__firstlineno__", "__static_attributes__"}
if set(vars(value)) - allowed:
raise ValueError("unknown imported class attributes")
expected_bases = tuple(vars(module).get(base.id) for base in node.bases if isinstance(base, ast.Name))
if not expected_bases:
expected_bases = (object,)
if value.__bases__ != expected_bases:
raise ValueError("unknown imported class bases")
for member, definition in methods.items():
_check_function(vars(value).get(member), definition, codes[name + "." + member], vars(module))
def _audit_provider(value):
cls = type(value)
module = sys.modules.get(cls.__module__)
if (module is None or vars(module).get("_PreviewOverrideWrapper") is not cls
or cls.__name__ != "_PreviewOverrideWrapper"):
raise ValueError("unrecognized Preview Override provider")
namespace = vars(module)
profile, source = _provider_source(namespace.get("__file__"))
codes, functions, classes = _compiled(source)
for name, node in functions.items():
_check_function(namespace.get(name), node, codes[name], namespace)
for name in ("_PreviewOverrideWrapper", "_AsyncPreviewEncoder"):
_check_plain_class(namespace.get(name), classes[name], codes, namespace,
sentinel=name == "_AsyncPreviewEncoder")
for alias, name in (("torch", "torch"), ("np", "numpy"), ("queue", "queue"),
("threading", "threading"), ("time", "time"), ("Image", "PIL.Image"),
("ImageOps", "PIL.ImageOps")):
if namespace.get(alias) is not sys.modules.get(name) or namespace.get(alias) is None:
raise ValueError("substituted preview module binding")
# Imported KJ bindings are canonical objects from the audited sibling files.
# No importlib/exec is used here; absent or substituted modules fail closed.
package = module.__name__.rsplit(".", 1)[0]
for relative, bindings in profile["siblings"].items():
sibling = sys.modules.get(package + "." + relative)
if sibling is None:
# ltxv_nodes is an explicitly optional import in the audited source.
if relative == "ltxv_nodes" and all(namespace.get(alias) is None for alias in bindings):
continue
raise ValueError("provider dependency not loaded")
dependency_source = _source(vars(sibling).get("__file__"), profile["sibling_sha256"][relative])
_check_selected(sibling, dependency_source, bindings.values())
for alias, original in bindings.items():
if namespace.get(alias) is not vars(sibling).get(original):
raise ValueError("substituted provider dependency")
# KJ reaches Core's preview factory directly. Unknown get_previewer hooks
# and an active Core approximate VAE stay outside this Phase A profile.
core = sys.modules.get("latent_preview")
if namespace.get("latent_preview") is not core or core is None:
raise ValueError("unknown Core preview module")
core_source = _source(vars(core).get("__file__"), profile["core_preview_sha256"])
_check_selected(core, core_source, ("get_previewer", "preview_to_image", "LatentPreviewer", "Latent2RGBPreviewer"))
patcher = sys.modules.get("comfy.patcher_extension")
if patcher is None:
raise ValueError("Core wrapper dispatcher not loaded")
dispatch_source = _source(vars(patcher).get("__file__"), profile["core_dispatch_sha256"])
_check_selected(patcher, dispatch_source, ("get_all_wrappers",))
method = getattr(vars(core).get("args"), "preview_method", None)
if (type(method) is not vars(core).get("LatentPreviewMethod")
or getattr(method, "name", None) not in ("NoPreviews", "Auto", "Latent2RGB")):
raise ValueError("Core approximate VAE is outside Phase A")
size = vars(core).get("MAX_PREVIEW_RESOLUTION")
if type(size) is not int or size <= 0:
raise ValueError("unknown Core preview setting")
return profile, {"method": method.name, "max_resolution": size}
def preview_override_signature(value):
"""Return a finite descriptor, never relaxing an unknown provider's guard."""
try:
profile, core_preview = _audit_provider(value)
state = vars(value)
if set(state) != _FIELDS or type(state["frames"]) is not list:
raise ValueError("unknown wrapper state")
if (state["vae"] is not None or state["audio_vae"] is not None
or type(state["tiny_vae"]) is not str or state["tiny_vae"] != "none"):
raise ValueError("VAE preview is outside Phase A")
for name, low, high in (("max_resolution", 0, 8192), ("jpeg_quality", 30, 100),
("preview_frames", 1, 1024), ("preview_fps", 1, 60)):
if type(state[name]) is not int or not low <= state[name] <= high:
raise ValueError("invalid fixed preview setting")
if (type(state["suppress_default"]) is not bool
or (state["node_id"] is not None and type(state["node_id"]) is not str)):
raise ValueError("invalid fixed preview setting")
settings = {key: state[key] for key in sorted(_FIELDS - {"frames", "vae", "audio_vae"})}
return {"adapter": "kj_preview_override_v1", "profile": profile["id"],
"profile_sha256": hashlib.sha256(json.dumps(profile, sort_keys=True, separators=(",", ":")).encode()).hexdigest(),
"source_sha256": profile["source_sha256"], "settings": settings,
"core_preview": core_preview}, True
except Exception as exc:
return {"adapter": "kj_preview_override_v1", "unsupported": str(exc)}, False
def preview_override_reuse_reasons(observed):
"""Classify the already observed KJ slot; never inspect MODEL a second time."""
wrappers = observed.get("wrappers") if type(observed) is dict else None
outer = wrappers.get("outer_sample") if type(wrappers) is dict else None
items = outer.get("kj_preview_override") if type(outer) is dict else None
if type(items) is not list:
return []
reasons = []
for item in items[:4096]:
if (type(item) is dict and item.get("adapter") == "kj_preview_override_v1"
and "unsupported" in item):
failure = item["unsupported"]
reason = _PREVIEW_REASONS.get(failure, _UNKNOWN_PREVIEW_REASON) if type(failure) is str else _UNKNOWN_PREVIEW_REASON
if reason not in reasons:
reasons.append(reason)
return reasons
def observe_model_wrappers(wrappers, observe):
"""Keep the no-KJ path exact; encode Core's key/list execution order for KJ."""
outer = wrappers.get("outer_sample") if type(wrappers) is dict else None
if not isinstance(outer, dict) or "kj_preview_override" not in outer:
return observe(wrappers)
if type(outer) is not dict or any(type(key) is not str for key in outer):
return {"unsupported": "unknown outer_sample container"}, False
if (len(wrappers) > 4096 or len(outer) > 4096
or any(type(key) is not str for key in wrappers)):
return {"unsupported": "wrapper container exceeds observation bounds"}, False
items = outer["kj_preview_override"]
if type(items) is not list or not items or len(items) > 4096:
return {"unsupported": "unknown kj_preview_override container"}, False
# A local copy holds placeholders while the ordinary observer checks all
# unrelated wrappers at their original nesting depth. Never mutate MODEL.
copied = dict(wrappers)
copied["outer_sample"] = dict(outer)
copied["outer_sample"]["kj_preview_override"] = [None] * len(items)
result, safe = observe(copied)
adapted = []
for value in items:
descriptor, recognized = preview_override_signature(value)
adapted.append(descriptor)
safe = safe and recognized
result["outer_sample"]["kj_preview_override"] = adapted
# Generic dict observation sorts keys. Keep a separate ordered sequence so
# cross-key execution order cannot disappear in canonical JSON hashing.
return {"wrappers": result, "outer_sample_order": list(outer)}, safe
-227
View File
@@ -1,227 +0,0 @@
"""UI-backed sequence prompt construction for H3 Continuum."""
from __future__ import annotations
import json
import re
from typing import Any, Mapping
from .v3.reference_images_v39 import (
REFERENCE_IMAGES_V39_TYPE,
ReferenceImagesV39Bundle,
)
MAX_CLIPS = 16
DEFAULT_CLIP_SECONDS = 4.0
def _number(value: Any, name: str, *, minimum: float = 0.0) -> float:
if isinstance(value, bool):
raise ValueError(f"{name} must be a number")
try:
result = float(value)
except (TypeError, ValueError) as exc:
raise ValueError(f"{name} must be a number") from exc
if result < minimum:
raise ValueError(f"{name} must be >= {minimum}")
return result
def default_sequence(clip_count: int = 2, clip_seconds: float = DEFAULT_CLIP_SECONDS) -> dict[str, Any]:
"""Return a portable editor state with one slot per clip."""
count = int(clip_count)
if count < 1 or count > MAX_CLIPS:
raise ValueError(f"clip_count must be between 1 and {MAX_CLIPS}")
duration = _number(clip_seconds, "clip_seconds", minimum=0.1)
return {
"version": 1,
"clips": [
{
"duration": duration,
"slots": [{"start": 0.0, "end": duration, "text": ""}],
}
for _ in range(count)
],
}
def normalize_sequence(value: Any, *, clip_count: int, default_seconds: float) -> dict[str, Any]:
"""Validate and normalize editor JSON without changing user-authored text."""
if isinstance(value, str):
try:
value = json.loads(value or "{}")
except json.JSONDecodeError as exc:
raise ValueError(f"timeline_json is not valid JSON: {exc.msg}") from exc
if not isinstance(value, Mapping):
raise ValueError("timeline_json must be a JSON object")
raw_clips = value.get("clips", [])
if not isinstance(raw_clips, list):
raise ValueError("timeline_json.clips must be a list")
count = int(clip_count)
if count < 1 or count > MAX_CLIPS:
raise ValueError(f"clip_count must be between 1 and {MAX_CLIPS}")
fallback_duration = _number(default_seconds, "default_chunk_seconds", minimum=0.1)
clips: list[dict[str, Any]] = []
for clip_index in range(count):
raw_clip = raw_clips[clip_index] if clip_index < len(raw_clips) else {}
if not isinstance(raw_clip, Mapping):
raise ValueError(f"clip {clip_index + 1} must be an object")
# Continuum projects use one chunk duration for every clip. Older
# editor state may still contain per-clip durations; ignore them so
# the single node control remains authoritative.
duration = fallback_duration
raw_slots = raw_clip.get("slots", [])
if not isinstance(raw_slots, list):
raise ValueError(f"clip {clip_index + 1} slots must be a list")
slots: list[dict[str, Any]] = []
for slot_index, raw_slot in enumerate(raw_slots):
if not isinstance(raw_slot, Mapping):
raise ValueError(f"clip {clip_index + 1} slot {slot_index + 1} must be an object")
start = _number(raw_slot.get("start", 0), f"clip {clip_index + 1} slot {slot_index + 1} start")
end = _number(raw_slot.get("end", duration), f"clip {clip_index + 1} slot {slot_index + 1} end")
if end <= start:
raise ValueError(f"clip {clip_index + 1} slot {slot_index + 1} end must be after start")
if end > duration + 1e-6:
raise ValueError(f"clip {clip_index + 1} slot {slot_index + 1} ends after the clip duration")
text = str(raw_slot.get("text", "")).strip()
if text:
slots.append({"start": start, "end": end, "text": text})
slots.sort(key=lambda slot: (slot["start"], slot["end"]))
if slots:
# The final shot owns the end of every uniform-duration chunk;
# there must be no unassigned tail on the editor track.
slots[-1]["end"] = duration
clips.append({"duration": duration, "slots": slots})
return {"version": 1, "clips": clips}
def format_sequence_prompt(
*,
shared_directive: str,
sequence: Mapping[str, Any],
soundscape: str = "",
music: str = "",
) -> str:
"""Format uniform clips as chunk blocks with relative shot timings.
Continuum treats square-bracket ranges as physical/logical chunk markers.
Shot timings belong inside each chunk and must therefore use plain
``start-end s:`` lines rather than creating additional Timeline sections.
"""
directive = str(shared_directive or "").strip()
sound = str(soundscape or "").strip() or "N/A"
music_text = str(music or "").strip() or "N/A"
sections: list[str] = []
cursor = 0.0
for clip in sequence["clips"]:
duration = float(clip["duration"])
chunk_lines = [f"[{cursor:g}-{cursor + duration:g}s]"]
if directive:
chunk_lines.append(directive)
for slot in clip["slots"]:
start = float(slot["start"])
end = float(slot["end"])
chunk_lines.append(f"{start:g}-{end:g}s: {slot['text']}")
chunk_lines.append(f"overall_soundscape: {sound}")
chunk_lines.append(f"non_diegetic_music: {music_text}")
sections.append("\n".join(chunk_lines))
cursor += duration
return "\n\n".join(sections).strip()
class DumasH3SequencePromptBuilder:
"""Build one Timeline prompt from clip tabs and clip-local time slots."""
DESCRIPTION = (
"DUMAS H3 prompt editor. Define clips and local time slots in the UI; "
"the node emits one absolute Timeline prompt for Sequence Prompt."
)
CATEGORY = "DUMAS H3/Prompt"
RETURN_TYPES = ("STRING", "INT", "FLOAT", REFERENCE_IMAGES_V39_TYPE)
RETURN_NAMES = ("sequence_prompt", "chunk_count", "chunk_duration", "reference_images")
FUNCTION = "build"
@classmethod
def INPUT_TYPES(cls):
state = json.dumps(default_sequence(), separators=(",", ":"))
return {
"required": {
"clip_count": ("INT", {"default": 2, "min": 1, "max": MAX_CLIPS, "step": 1}),
"default_chunk_seconds": ("FLOAT", {"default": DEFAULT_CLIP_SECONDS, "min": 0.1, "max": 30.0, "step": 0.1}),
"shared_directive": ("STRING", {"default": "", "multiline": True}),
"timeline_json": ("STRING", {"default": state, "multiline": True}),
"soundscape": ("STRING", {"default": "", "multiline": True}),
"music": ("STRING", {"default": "", "multiline": True}),
},
"optional": {
f"reference_image_{index}": ("IMAGE", {
"display_name": f"Reference Image {index} (@R{index})",
"tooltip": f"Fixed Reference Image {index}; use @R{index} in a clip prompt.",
}) for index in range(1, 10)
}
}
def build(
self,
clip_count,
default_chunk_seconds,
shared_directive,
timeline_json,
soundscape,
music,
**kwargs,
):
sequence = normalize_sequence(
timeline_json,
clip_count=int(clip_count),
default_seconds=float(default_chunk_seconds),
)
selectors = []
directive = str(shared_directive or "").strip()
sound = str(soundscape or "").strip() or "N/A"
music_text = str(music or "").strip() or "N/A"
for reference_index in range(1, 10):
chunks = []
for chunk_index, clip in enumerate(sequence["clips"], start=1):
# Match against the same per-chunk content emitted by
# format_sequence_prompt, not only the editable shot lines.
text = "\n".join([
directive,
*(str(slot.get("text", "")) for slot in clip["slots"]),
f"overall_soundscape: {sound}",
f"non_diegetic_music: {music_text}",
])
if kwargs.get(f"reference_image_{reference_index}") is not None and re.search(
rf"@(R{reference_index}|image{reference_index})(?![A-Za-z0-9_])",
text,
re.IGNORECASE,
):
chunks.append(chunk_index)
selectors.append(",".join(map(str, chunks)) or "off")
images = tuple(kwargs.get(f"reference_image_{index}") for index in range(1, 10))
reference_bundle = ReferenceImagesV39Bundle(
images=images,
reference_use="Per chunk",
selectors=tuple(selectors),
)
return (
format_sequence_prompt(
shared_directive=shared_directive,
sequence=sequence,
soundscape=soundscape,
music=music,
),
len(sequence["clips"]),
float(default_chunk_seconds),
reference_bundle,
)
NODE_CLASS_MAPPINGS = {"DumasH3SequencePromptBuilder": DumasH3SequencePromptBuilder}
NODE_DISPLAY_NAME_MAPPINGS = {
"DumasH3SequencePromptBuilder": "DUMAS H3 Sequence Prompt Builder",
}
+6 -6
View File
@@ -1,7 +1,7 @@
[project]
name = "comfyui-h3-continuum-dumas"
version = "3.9.4.post1"
description = "H3 Continuum fork with per-clip tiled latent upscaling"
name = "comfyui-h3-continuum"
version = "3.9.10"
description = "Integrated N-chunk continuation for ComfyUI native MiniMax H3 video and audio"
readme = "README.md"
requires-python = ">=3.10"
license = { file = "LICENSE" }
@@ -10,9 +10,9 @@ dependencies = []
keywords = ["comfyui", "minimax-h3", "video", "audio", "continuation"]
[project.urls]
Repository = "https://git.dumas.ddns.net/chris.dumas/ComfyUI-H3-Continuum-Tiled"
Repository = "https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum"
[tool.comfy]
PublisherId = "chris.dumas"
DisplayName = "DUMAS H3 Continuum Tiled"
PublisherId = "ukr8b3g-cmyk"
DisplayName = "H3 Continuum"
requires-comfyui = ">=0.32.0"
+430 -3
View File
@@ -38,6 +38,10 @@ from .branch_provenance import (
resolve_chain,
)
from .constants import CONTINUUM_ACTUAL_PREFIX_STEPS, continuity_storage_mode
from .spectrum_signature import spectrum_binding_signature
from .preview_override_signature import (
PREVIEW_REUSE_REASONS, observe_model_wrappers, preview_override_reuse_reasons,
)
from .v2.session import make_session, validate_chunk_entry
from .v3.review_control import (
GENERATION_MODE_REVIEW,
@@ -552,10 +556,404 @@ def _observe_value(value: Any, *, depth: int = 0) -> tuple[Any, bool]:
return {"type": _qualified(value), "repr": rendered[:256]}, False
# The generic observer remains byte-for-byte compatible with unaffected v6 Runs.
# Only the Core setter / registered PyTorch and Kitchen override gets this
# finite descriptor. Never import Core/Kitchen or execute an attention kernel.
def _attention_code(code):
constants = tuple(
_attention_code(item) if isinstance(item, type(code)) else item
for item in code.co_consts
)
return code.replace(co_consts=constants, co_filename="", co_firstlineno=0)
def _attention_class(value):
import dataclasses
methods = {}
safe = True
for name, method in vars(value).items():
if isinstance(method, property):
methods[name] = {"property_unobserved": True}
safe = False
continue
if not inspect.isfunction(method):
continue
references = {}
for cell_name, cell in zip(method.__code__.co_freevars, method.__closure__ or ()):
target = cell.cell_contents
origin = (isinstance(target, type) and dataclasses.is_dataclass(value)
and target is not value and _qualified(target) == _qualified(value)
and getattr(target, "__dataclass_fields__", None) is value.__dataclass_fields__
and getattr(target, "__init__", None) is value.__init__)
if target is value or origin or target is object or (
isinstance(target, type) and target.__module__ in {"builtins", "dataclasses"}):
references[cell_name] = target
descriptor, complete = _attention_function(method, references=references)
descriptor["reference_targets"] = {
k: {"type": _qualified(v), "owner": v is value,
"dataclass_origin": isinstance(v, type) and v is not value and _qualified(v) == _qualified(value)}
for k, v in references.items()
}
methods[name] = descriptor
safe = safe and complete
return {"class": _qualified(value), "slots": list(getattr(value, "__slots__", ())),
"methods": methods}, safe
def _attention_globals(function):
import builtins
import dis
import sys
from types import ModuleType
names = set()
def visit(code):
names.update(i.argval for i in dis.get_instructions(code) if i.opname == "LOAD_GLOBAL")
for child in code.co_consts:
if isinstance(child, type(code)):
visit(child)
visit(function.__code__)
result = {}
safe = True
for name in sorted(names):
if name not in function.__globals__:
if not hasattr(builtins, name):
safe = False
continue
value = function.__globals__[name]
if isinstance(value, ModuleType):
if sys.modules.get(value.__name__) is not value or not value.__name__.startswith(
("torch", "comfy", "math", "functools", "typing")):
return result, False
path = getattr(value, "__file__", None)
result[name] = {"module": value.__name__,
"file_sha256": _attention_source_file(path) if path else None}
elif inspect.isbuiltin(value):
owner = sys.modules.get(value.__module__)
if (owner is None or getattr(owner, value.__name__, None) is not value or
not value.__module__.startswith(("torch.", "comfy_kitchen.", "_thread", "math", "builtins"))):
return result, False
path = getattr(owner, "__file__", None)
result[name] = {"builtin": _qualified(value), "python_version": list(sys.version_info[:3]),
"file_sha256": _attention_source_file(path) if path else None}
elif inspect.isfunction(value):
observed, complete = _observe(value)
if not complete:
# The only global cycle admitted is another registered Core
# backend, used by Kitchen's low-precision fallback.
attention = sys.modules.get("comfy.ldm.modules.attention")
registry = getattr(attention, "REGISTERED_ATTENTION_FUNCTIONS", {})
backend = next((n for n in ("pytorch", "comfy_kitchen_int8")
if registry.get(n) is value), None)
cells = dict(zip(value.__code__.co_freevars,
(c.cell_contents for c in value.__closure__ or ())))
if (backend is None or set(cells) != {"func", "wrapper"} or
cells["wrapper"] is not value or value.__globals__ is not vars(attention)):
return result, False
wrapper, wrapper_safe = _attention_function(value, references=cells)
raw, raw_safe = _observe(cells["func"])
observed = {"backend": backend, "wrapper": wrapper, "raw": raw}
complete = wrapper_safe and raw_safe and value.container_function is None
attributes, attributes_safe = _observe(vars(value)) if complete and not (
isinstance(observed, dict) and "backend" in observed) else ({}, complete)
result[name] = {"function": observed, "attributes": attributes}
safe = safe and complete and attributes_safe
elif isinstance(value, type):
if not value.__module__.startswith(("torch", "comfy")):
return result, False
result[name], complete = _attention_class(value)
safe = safe and complete
else:
try:
result[name] = _attention_provider_state(value)
except Exception:
result[name] = {"type": _qualified(value), "unobservable": True}
safe = False
return result, safe
def _attention_function(function, *, references=None):
if not inspect.isfunction(function):
return _observe(function)[0], False
references = references or {}
closure = []
safe = True
for name, cell in zip(function.__code__.co_freevars, function.__closure__ or ()):
value = cell.cell_contents
if name in references and value is references[name]:
observed, complete = {"reference": name}, True
else:
observed, complete = _observe(value)
closure.append({"name": name, "value": observed})
safe = safe and complete
attributes = dict(vars(function))
for name, value in references.items():
if name == "func" and attributes.get("__wrapped__") is value:
attributes.pop("__wrapped__")
# Both container paths are recorded separately by the adapter.
attributes.pop("container_function", None)
defaults, defaults_safe = _observe(function.__defaults__)
kwdefaults, kw_safe = _observe(function.__kwdefaults__)
attributes, attributes_safe = _observe(attributes)
globals_value, globals_safe = _attention_globals(function)
return {
"callable": _qualified(function),
"code_sha256": hashlib.sha256(marshal.dumps(_attention_code(function.__code__))).hexdigest(),
"defaults": defaults, "kwdefaults": kwdefaults,
"closure": closure, "attributes": attributes, "globals": globals_value,
}, safe and defaults_safe and kw_safe and attributes_safe and globals_safe
@functools.lru_cache(maxsize=512)
def _attention_file_hash(path, size, mtime_ns, ctime_ns):
return _file_sha256(Path(path))
def _attention_source_file(path):
path = Path(path)
stat = path.stat()
return _attention_file_hash(str(path), stat.st_size, stat.st_mtime_ns, stat.st_ctime_ns)
def _attention_provider_state(value, *, depth=0):
# Provider constraints contain torch dtype / SDP enums, not MODEL tensors.
if depth > 10:
raise ValueError("provider state exceeds observation depth")
import dataclasses
for name in ("MISSING", "_FIELD_CLASSVAR", "_FIELD_INITVAR", "_HAS_DEFAULT_FACTORY"):
if value is getattr(dataclasses, name):
return {"dataclass_sentinel": name}
if isinstance(value, torch.dtype):
return {"torch_dtype": str(value)}
if value is None or isinstance(value, (bool, int, float, str)):
return value
if isinstance(value, (dict, tuple, list, set, frozenset)) and len(value) > _OBSERVATION_MAX_ITEMS:
raise ValueError("provider state exceeds item bound")
if isinstance(value, (tuple, list)):
return [_attention_provider_state(v, depth=depth + 1) for v in value]
if isinstance(value, (set, frozenset)):
return sorted((_attention_provider_state(v, depth=depth + 1) for v in value), key=_canonical)
if isinstance(value, dict):
return sorted([
[_attention_provider_state(k, depth=depth + 1),
_attention_provider_state(v, depth=depth + 1)]
for k, v in value.items()
], key=_canonical)
import dataclasses
if dataclasses.is_dataclass(value) and not isinstance(value, type):
return {"type": _qualified(value), "fields": {
field.name: _attention_provider_state(getattr(value, field.name), depth=depth + 1)
for field in dataclasses.fields(value)
}}
sdp = getattr(getattr(torch.nn, "attention", None), "SDPBackend", None)
if sdp is not None and isinstance(value, sdp):
return {"torch_sdp_backend": int(value.value)}
if inspect.isfunction(value):
result, safe = _attention_function(value)
if safe:
return result
raise ValueError("unsupported attention provider state")
def _attention_providers(attention, backend):
import sys
import importlib.metadata
ops = sys.modules.get("comfy.ops")
if ops is None:
raise ValueError("Core attention providers are not loaded")
functions = {
"reshape": attention._reshape_qkv_to_heads,
"heads": attention._heads_from_dim,
"sdpa": ops.scaled_dot_product_attention,
"repeat_kv": ops.repeat_kv_for_gqa,
}
state = {"batch_limit": attention.SDP_BATCH_LIMIT}
for name in ("SDPA_BACKEND_PRIORITY", "SDPA_CONTIGUOUS_KV_L2_SIZE"):
if hasattr(ops, name):
state[name] = getattr(ops, name)
container = attention.AttentionTensorContainer
functions.update({
"container_init": container.__init__,
"container_peek": container.peek, "container_take": container.take,
})
# The provider's native binding must still be the loaded PyTorch binding.
if torch.nn.functional.scaled_dot_product_attention is not torch._C._nn.scaled_dot_product_attention:
raise ValueError("unsupported PyTorch SDPA replacement")
cuda = torch.backends.cuda
state["torch_sdpa"] = {
name: getattr(cuda, name)() for name in
("flash_sdp_enabled", "mem_efficient_sdp_enabled", "math_sdp_enabled",
"cudnn_sdp_enabled", "fp16_bf16_reduction_math_sdp_allowed") if hasattr(cuda, name)
}
state["torch_matmul"] = {
"allow_tf32": cuda.matmul.allow_tf32,
"float32_precision": torch.get_float32_matmul_precision(),
}
files = {
"core_attention": _attention_source_file(attention.__file__),
"core_ops": _attention_source_file(ops.__file__),
"torch_functional": _attention_source_file(torch.nn.functional.__file__),
"torch_native": _attention_source_file(torch._C.__file__),
}
versions = {"torch": str(torch.__version__), "torch_git": str(torch.version.git_version)}
if backend == "comfy_kitchen_int8":
kitchen = sys.modules.get("comfy_kitchen")
if kitchen is None or getattr(attention, "comfy_kitchen", None) is not kitchen:
raise ValueError("Kitchen provider is not loaded")
functions["int8_inputs"] = attention._comfy_kitchen_int8_inputs
for name in ("int8_attention", "prequantize_int8_attention", "int8_attention_from_prequantized"):
functions[name] = getattr(kitchen, name)
versions["comfy_kitchen"] = importlib.metadata.version("comfy-kitchen")
root = Path(kitchen.__file__).parent
paths = sorted(p for p in root.rglob("*") if p.is_file() and p.suffix in {".py", ".pyd", ".so"})
if len(paths) > _OBSERVATION_MAX_ITEMS:
raise ValueError("Kitchen provider exceeds file observation bound")
files["comfy_kitchen"] = {p.relative_to(root).as_posix(): _attention_source_file(p) for p in paths}
# Capture runtime replacement of Python numerical helpers too.
module = sys.modules.get("comfy_kitchen.sage_attention")
if module is None:
raise ValueError("Kitchen implementation is not loaded")
functions.update({
"kitchen_helper." + name: value for name, value in vars(module).items()
if inspect.isfunction(value) and value.__module__ == module.__name__
})
# Kitchen INT8 dispatches directly to its CUDA/HIP extension, not the
# general op registry. Record only the dispatch actually used here.
state["kitchen_constants"] = {
name: value for name, value in vars(module).items()
if name.isupper() and not name.startswith("__")
}
for slot in ("_cuda_backend", "_hip_backend"):
provider = getattr(module, slot)
if provider is None:
state[slot] = None
continue
if sys.modules.get(provider.__name__) is not provider:
raise ValueError("unregistered Kitchen native provider")
native = getattr(provider, "_C", None)
if native is not None and sys.modules.get(native.__name__) is not native:
raise ValueError("unsupported Kitchen native binding")
state[slot] = {
"module": provider.__name__,
"extension_available": getattr(provider, "_EXT_AVAILABLE", None),
"native_sha256": _attention_source_file(native.__file__) if native is not None else None,
}
functions.update({
slot + "." + name: function for name, function in vars(provider).items()
if inspect.isfunction(function) and
(name.startswith(("sage_", "_sage_")) or name in {"_dl", "_wrap_for_dlpack"})
})
observed_functions = {}
for name, function in functions.items():
owner = sys.modules.get(getattr(function, "__module__", ""))
if (not inspect.isfunction(function) or owner is None or
not function.__module__.startswith(("comfy.", "comfy_kitchen.")) or
function.__globals__ is not vars(owner)):
raise ValueError("unsupported attention helper binding: " + name)
descriptor, safe = _attention_function(function)
if not safe:
raise ValueError("unobservable attention helper: " + name)
observed_functions[name] = descriptor
return {"versions": versions, "files": files, "functions": observed_functions,
"state": _attention_provider_state(state)}
def _core_attention_override(value):
import sys
attention = sys.modules.get("comfy.ldm.modules.attention")
patcher = sys.modules.get("comfy.model_patcher")
if attention is None or patcher is None or not inspect.isfunction(value):
return None
setter = patcher.ModelPatcher.set_model_optimized_attention
override_codes = [c for c in setter.__code__.co_consts
if isinstance(c, type(setter.__code__)) and c.co_name == "optimized_attention_override"]
wrapper_codes = [c for c in attention.wrap_attn.__code__.co_consts
if isinstance(c, type(setter.__code__)) and c.co_name == "wrapper"]
if len(override_codes) != 1 or len(wrapper_codes) != 1 or value.__code__ != override_codes[0]:
return None
cells = dict(zip(value.__code__.co_freevars, (c.cell_contents for c in value.__closure__ or ())))
if set(cells) != {"optimized_attention"}:
return None
wrapped = cells["optimized_attention"]
registry = attention.REGISTERED_ATTENTION_FUNCTIONS
backend = next((name for name in ("pytorch", "comfy_kitchen_int8")
if registry.get(name) is wrapped and
getattr(attention, "attention_" + name, None) is wrapped), None)
if backend is None or not inspect.isfunction(wrapped) or wrapped.__code__ != wrapper_codes[0]:
return None
cells = dict(zip(wrapped.__code__.co_freevars, (c.cell_contents for c in wrapped.__closure__ or ())))
if set(cells) != {"func", "wrapper"} or cells["wrapper"] is not wrapped or getattr(wrapped, "__wrapped__", None) is not cells["func"]:
return None
raw = cells["func"]
if (value.__globals__ is not vars(patcher) or wrapped.__globals__ is not vars(attention)
or not inspect.isfunction(raw) or raw.__globals__ is not vars(attention)):
return None
expected_container = (getattr(attention, "_attention_comfy_kitchen_int8_containers", None)
if backend == "comfy_kitchen_int8" else None)
inner = getattr(wrapped, "container_function", None)
outer = getattr(value, "container_function", None)
components = {}
safe = inner is expected_container and outer is expected_container
if expected_container is not None:
safe = safe and inspect.isfunction(expected_container) and expected_container.__globals__ is vars(attention)
for name, function, refs in (
("override", value, {"optimized_attention": wrapped}),
("wrapper", wrapped, {"wrapper": wrapped, "func": raw}),
("raw", raw, {}),
("inner_container", inner, {}),
("outer_container", outer, {}),
):
if function is None:
descriptor, complete = None, True
else:
descriptor, complete = _attention_function(function, references=refs)
components[name] = descriptor
safe = safe and complete
result = {"core_attention_signature_version": 1, "backend": backend, **components,
"outer_container_present": hasattr(value, "container_function")}
try:
result["providers"] = _attention_providers(attention, backend)
except Exception as exc:
result["observation_error"] = type(exc).__name__
result["reuse_disabled_reason"] = "Core attention provider implementation/state is unobservable"
safe = False
return result, safe
def _observe_model_options(options):
spectrum = None
if "spectrum_h3_binding" in options:
spectrum = spectrum_binding_signature(options["spectrum_h3_binding"])
options = dict(options)
options["spectrum_h3_binding"] = None
observed, safe = _observe_attention_model_options(options)
if spectrum is not None:
observed["spectrum_h3_binding"] = spectrum[0]
safe = safe and spectrum[1]
return observed, safe
def _observe_attention_model_options(options):
transformer = options.get("transformer_options")
if not isinstance(transformer, dict) or "optimized_attention_override" not in transformer:
return _observe(options)
try:
adapted = _core_attention_override(transformer["optimized_attention_override"])
except Exception:
adapted = None
if adapted is None:
# Unknown overrides keep their prior descriptor and safe-side policy.
return _observe(options)
copied = dict(options)
copied["transformer_options"] = dict(transformer)
copied["transformer_options"]["optimized_attention_override"] = None
observed, safe = _observe(copied)
if not isinstance(observed.get("transformer_options"), dict) or "truncated" in observed["transformer_options"]:
return observed, False
observed["transformer_options"]["optimized_attention_override"] = adapted[0]
return observed, safe and adapted[1]
def _model_signature(model: Any, legacy_fingerprint: str) -> tuple[dict[str, Any], bool]:
base = getattr(model, "model", None)
inner = getattr(base, "diffusion_model", None)
wrappers, wrappers_safe = _observe(getattr(model, "wrappers", {}) or {})
wrappers, wrappers_safe = observe_model_wrappers(getattr(model, "wrappers", {}) or {}, _observe)
patches, patches_safe = _observe(getattr(model, "patches", {}) or {})
options = getattr(model, "model_options", {}) or {}
if not isinstance(options, dict):
@@ -565,7 +963,7 @@ def _model_signature(model: Any, legacy_fingerprint: str) -> tuple[dict[str, Any
transformer.pop("h3_continuum_join_context", None)
if "transformer_options" in observed_options:
observed_options["transformer_options"] = transformer
observed_options, options_safe = _observe(observed_options)
observed_options, options_safe = _observe_model_options(observed_options)
weights: list[dict[str, Any]] = []
weights_safe = True
try:
@@ -633,6 +1031,31 @@ def _reusable_manifest(manifest: dict[str, Any]) -> bool:
)
_UNKNOWN_REUSE_REASON = "Saved generation conditions could not be fully verified; see Detailed Report"
_PUBLIC_REUSE_REASONS = PREVIEW_REUSE_REASONS | frozenset((
"MODEL wrapper/patch contract is not completely observable",
"SAMPLER function/options are not completely observable",
"CLIP/Qwen encoder contract is not completely observable",
"Video VAE contract is not completely observable",
"Reference Audio VAE contract is not completely observable",
"Driving Audio VAE contract is not completely observable",
_UNKNOWN_REUSE_REASON,
))
def _project_resume_diagnostics(manifest: dict[str, Any]) -> dict[str, Any]:
"""Expose bounded saved reasons without changing contracts or old manifests."""
source = manifest.get("resume_disabled_reasons")
if _reusable_manifest(manifest) or type(source) is not list:
return {}
reasons = []
for value in source[:8]:
reason = value if type(value) is str and value in _PUBLIC_REUSE_REASONS else _UNKNOWN_REUSE_REASON
if reason not in reasons:
reasons.append(reason[:256])
return {"resume_disabled_reasons": reasons} if reasons else {}
def _module_parameter_dtype(module: Any, name: str, parameter: torch.Tensor) -> torch.dtype:
"""Observe ordinary weights at Core's declared load dtype without casting them."""
# Quantized/custom Tensor representations retain their original fingerprint.
@@ -957,6 +1380,7 @@ def build_sampling_contract(
raise RunStorageError("routed Reference plan cannot use legacy Reference contract")
last_frame_hash = _canonical_optional_hash(last_frame_hash)
model_value, model_safe = _signature_observation(_model_signature, model, model_fingerprint_value)
preview_reasons = preview_override_reuse_reasons(model_value.get("wrappers"))
clip_value, clip_safe = _clip_signature(clip)
has_first = str(first_frame_hash).lower() not in {"", "none", "null"}
has_last = str(last_frame_hash).lower() not in {"", "none", "null"}
@@ -1127,6 +1551,7 @@ def build_sampling_contract(
reasons = list(upstream_graph_reasons or []) if graph_authoritative else []
if not model_safe:
reasons.append("MODEL wrapper/patch contract is not completely observable")
reasons.extend(preview_reasons)
if not sampler_safe:
reasons.append("SAMPLER function/options are not completely observable")
if not clip_safe:
@@ -1663,6 +2088,7 @@ class RunStorageController:
branch_regenerate_from=value.get("branch_regenerate_from"),
active_chain=chains.get(revision_id, []),
)
revision.update(_project_resume_diagnostics(value))
revisions.append(revision)
canonical_storage_revision_id, canonical_chain = self._canonical_project_selection(
catalog=catalog,
@@ -1688,7 +2114,8 @@ class RunStorageController:
"group_revisions": sorted(
[dict(item,
sampling_contract_version=(manifests.get(str(item.get("storage_revision_id"))) or {}).get("sampling_contract_version"),
generation_reusable=_reusable_manifest(manifests.get(str(item.get("storage_revision_id"))) or {}))
generation_reusable=_reusable_manifest(manifests.get(str(item.get("storage_revision_id"))) or {}),
**_project_resume_diagnostics(manifests.get(str(item.get("storage_revision_id"))) or {}))
for item in catalog.values()],
key=lambda item: (
int(item["group"]["physical_group"]),
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
+1 -1
View File
@@ -49,7 +49,7 @@ def review_queue_results():
capture_output=True, text=True, encoding="utf-8", timeout=45,
)
records = json.loads(result.stdout)
assert len(records) == 62, "The complete queue fixture must finish, including Workflow rehydration and v5 history"
assert len(records) == 90, "The complete queue fixture must finish, including rehydration, v5 history and read-only reasons"
failures = [item for item in records if not item["pass"]]
assert result.returncode == 0 and not failures, result.stdout + result.stderr
return {item["name"]: item for item in records}
File diff suppressed because one or more lines are too long
+19
View File
@@ -0,0 +1,19 @@
def _observe_model_options(options):
transformer = options.get("transformer_options")
if not isinstance(transformer, dict) or "optimized_attention_override" not in transformer:
return _observe(options)
try:
adapted = _core_attention_override(transformer["optimized_attention_override"])
except Exception:
adapted = None
if adapted is None:
# Unknown overrides keep their prior descriptor and safe-side policy.
return _observe(options)
copied = dict(options)
copied["transformer_options"] = dict(transformer)
copied["transformer_options"]["optimized_attention_override"] = None
observed, safe = _observe(copied)
if not isinstance(observed.get("transformer_options"), dict) or "truncated" in observed["transformer_options"]:
return observed, False
observed["transformer_options"]["optimized_attention_override"] = adapted[0]
return observed, safe and adapted[1]
+111
View File
@@ -193,4 +193,115 @@ oldSampler.inputs[0].link = null;
assert.equal(pruneV39LegacyReferenceInputs(oldSampler), 1);
assert.deepEqual(oldSampler.inputs.map((input) => input.name), ['reference_images']);
assert.equal(pruneV39LegacyReferenceInputs({ ...oldSampler, comfyClass: 'H3ContinuumSamplerV38' }), 0);
// External INT values are not frontend widget values. Exercise the production
// helper in fresh graphs without resolving or guessing any upstream value.
function assignmentFixture({ value = 3, external = false, link = 90, saved = 'off', missing = false, other = null } = {}) {
const graph = { links: {}, nodes: new Map(), getNodeById(id) { return this.nodes.get(id); } };
const calls = { native: 0, notify: 0, size: 0 };
const controls = Array.from({ length: 9 }, (_, i) => ({ name: `reference_r${i + 1}_chunks`,
value: i === 0 ? saved : 'off', options: {}, callback() { calls.native++; } }));
const helper = { id: 1, graph, comfyClass: 'H3ContinuumReferenceImagesV39', size: [440, 500],
widgets: [{ name: 'reference_use', value: 'Per chunk' }, ...controls],
inputs: [{ name: 'reference_image_1', link: 100 }], outputs: [{ name: 'reference_images', links: [10] }],
addDOMWidget(_name, _type, host) { this.panel = host.children[0]; this.display = { options: {} }; return this.display; },
computeSize() { return [440, 310 + this.display.computeSize(440)[1]]; },
setSize(value) { calls.size++; this.size = value; }, setDirtyCanvas() {},
};
const sampler = { id: 2, graph, comfyClass: 'H3ContinuumSamplerV39',
widgets: missing ? [] : [{ name: 'chunks', value }],
inputs: [{ name: 'reference_images', link: 10 }, { name: 'chunks', link: external ? link : null }],
__h3ContinuumIntuitiveUxRefresh() { calls.notify++; }, setDirtyCanvas() {},
};
graph.nodes.set(1, helper); graph.nodes.set(2, sampler); graph.links[10] = { origin_id: 1, target_id: 2 };
if (other !== null) {
const second = { ...sampler, id: 3, widgets: [{ name: 'chunks', value: other }],
inputs: [{ name: 'reference_images', link: 11 }, { name: 'chunks', link: null }] };
graph.nodes.set(3, second); graph.links[11] = { origin_id: 1, target_id: 3 }; helper.outputs[0].links.push(11);
}
configureV39ReferenceImages(helper);
return { graph, helper, sampler, calls, control: controls[0],
text() { const values = []; const collect = (node) => { values.push(node.textContent || ''); (node.children || []).forEach(collect); }; collect(helper.panel); return values.join('\n'); },
checkbox(n) { return visit(helper.panel, (item) => item['aria-label'] === `Image 1, Chunk ${n}`); },
action(label) { const button = visit(helper.panel, (item) => item.tag === 'button' && item.textContent === label); assert(button && !button.disabled, label); button.onclick(); },
refresh() { refreshV39ReferenceImagesForSampler(sampler); },
};
}
const externalCases = [];
function externalCase(name, run) {
try { run(); externalCases.push({ name, pass: true }); }
catch (error) { externalCases.push({ name, pass: false, error: error.message }); }
}
function unknown(f, reason) {
assert.match(f.text(), /range 1–16/);
assert.match(f.text(), reason);
assert(f.checkbox(3));
assert(!f.text().includes('saved beyond range'));
}
externalCase('J01 direct three keeps native range', () => {
const f = assignmentFixture(); assert.match(f.text(), /3 chunks/); assert(f.checkbox(3)); assert(!f.checkbox(4));
f.action('All'); assert.equal(f.control.value, '1,2,3'); f.action('Off'); assert.equal(f.control.value, 'off');
});
externalCase('J02 external three and widget two allow C3', () => unknown(assignmentFixture({ value: 2, external: true }), /connected node/));
externalCase('J03 external with matching widget remains unresolved', () => unknown(assignmentFixture({ external: true }), /connected node/));
externalCase('J04 link zero is external', () => unknown(assignmentFixture({ external: true, link: 0 }), /connected node/));
externalCase('J05 unrestored link stays external', () => unknown(assignmentFixture({ external: true, link: 99999 }), /connected node/));
externalCase('J06 disconnected or absent socket uses direct value', () => {
const f = assignmentFixture(); f.sampler.inputs.pop(); f.refresh(); assert.match(f.text(), /3 chunks/);
});
externalCase('J07 unavailable or invalid widget preserves selection', () => {
for (const options of [{ missing: true }, { value: 0 }, { value: 17 }, { value: 2.5 }, { value: 'bad' }]) {
const f = assignmentFixture({ ...options, saved: '1,2,3' }); unknown(f, /not available/); assert.equal(f.control.value, '1,2,3');
}
});
externalCase('J08 multiple direct three targets agree', () => assert.match(assignmentFixture({ other: 3 }).text(), /3 chunks/));
externalCase('J09 multiple different direct counts stay unresolved', () => unknown(assignmentFixture({ other: 2 }), /different chunk counts/));
externalCase('J10 any externally connected target makes range unresolved', () => unknown(assignmentFixture({ external: true, other: 3 }), /connected node/));
externalCase('J11 unknown reason refreshes even when range is unchanged', () => {
const f = assignmentFixture({ external: true }); unknown(f, /connected node/);
f.helper.outputs[0].links = []; f.helper.onConnectionsChange(); unknown(f, /Connect this helper to a Sampler/);
assert(!f.text().includes('connected node'));
});
externalCase('J12 upstream value changes do not rewrite or rebuild', () => {
const f = assignmentFixture({ external: true, saved: '1,2' }); const first = f.helper.panel.children[0];
const counts = { ...f.calls }; f.graph.nodes.set(90, { value: 4 }); f.refresh();
assert.equal(f.helper.panel.children[0], first); assert.equal(f.control.value, '1,2'); assert.deepEqual(f.calls, counts);
});
externalCase('J13 saved C3 remains visible and is not called outside range', () => {
const f = assignmentFixture({ value: 2, external: true, saved: '1,2,3' }); f.refresh();
assert.equal(f.control.value, '1,2,3'); assert(f.checkbox(3).checked); assert.match(f.text(), /Saved assignments: 1, 2, 3/);
});
externalCase('J14 old one-two is never automatically expanded', () => {
const f = assignmentFixture({ value: 2, external: true, saved: '1,2' }); configureV39ReferenceImages(f.helper); f.refresh();
assert.equal(f.control.value, '1,2'); assert(!f.checkbox(3).checked);
});
externalCase('J15 unknown All operates over all pages', () => {
const f = assignmentFixture({ external: true }); f.action('▶'); f.action('▶'); f.action('All 1–16');
assert.equal(f.control.value, Array.from({ length: 16 }, (_, i) => i + 1).join(','));
});
externalCase('J16 unknown Off clears future saved assignments', () => {
const f = assignmentFixture({ external: true, saved: '1,2,3,16' }); f.action('Off 1–16'); assert.equal(f.control.value, 'off');
});
externalCase('J17 direct All and Off preserve saved future six', () => {
const f = assignmentFixture({ saved: '6' }); f.action('All'); assert.equal(f.control.value, '1,2,3,6'); f.action('Off'); assert.equal(f.control.value, '6');
});
externalCase('J18 C3 updates native value and notifies once', () => {
const f = assignmentFixture({ value: 2, external: true }); const counts = { ...f.calls }; const box = f.checkbox(3);
assert(box); box.checked = true; box.onchange(); assert.equal(f.control.value, '3');
assert.equal(f.calls.native, counts.native + 1); assert.equal(f.calls.notify, counts.notify + 1); assert.equal(f.calls.size, counts.size);
});
externalCase('J19 independent helpers and multiple targets stay independent', () => {
const f = assignmentFixture({ external: true, other: 3 }), g = assignmentFixture({ external: true });
f.action('All 1–16'); assert.equal(f.calls.notify, 2); assert.equal(g.control.value, 'off'); assert.equal(g.calls.notify, 0);
});
externalCase('J20 repeated configure keeps callbacks layout and width', () => {
const f = assignmentFixture({ external: true }); const callback = f.control.callback, sizes = f.calls.size;
for (let i = 0; i < 100; i++) configureV39ReferenceImages(f.helper);
assert.equal(f.control.callback, callback); assert.equal(f.calls.size, sizes); assert.equal(f.helper.size[0], 440);
});
externalCase('J21 malformed saved selector remains disabled and unchanged', () => {
const f = assignmentFixture({ external: true, saved: 'invalid' }); assert.equal(f.control.value, 'invalid');
assert(f.checkbox(3).disabled); assert(visit(f.helper.panel, (item) => item.textContent === 'All 1–16').disabled);
});
console.log(JSON.stringify(externalCases));
assert.equal(externalCases.filter((item) => !item.pass).length, 0, 'external Chunk assignment contracts failed');
console.log('V3.9 Reference Images UI contract PASS');
+120
View File
@@ -697,5 +697,125 @@ await test('archived v5 lineage remains browsable after a v6 canonical run',asyn
const n=e.makeNode();await e.load(n,current);
assert.equal(e.f.takeCatalog(n).length,3);
});
await test('v6 read-only reports saved preview source and keeps actions disabled',async e=>{
const p=project(1),reason='KJ/Core preview implementation is outside the audited source profiles';
for(const item of [...p.group_revisions,...p.revisions]){
item.generation_reusable=false;item.resume_disabled_reasons=[reason];
}
const n=e.makeNode();e.w(n,'generation_mode').value='Review Each Chunk';await e.load(n,p);
assert.match(e.f.reviewStatus(n),/Sampling v6 Take is read-only/);
assert.match(e.f.reviewStatus(n),/Bypass KJ Model Preview Override/);
assert(!e.visible(n,'Use it and continue'));
const before=[e.w(n,'take_action').value,e.w(n,'take_revision_id').value,e.w(n,'run_name').value];
e.f.selectTakeAction(n,'Use This Take');e.f.selectTakeAction(n,'Continue From Here');
assert.deepEqual([e.w(n,'take_action').value,e.w(n,'take_revision_id').value,e.w(n,'run_name').value],before);
});
await test('preview setting remedies preserve generation VAE guidance',async e=>{
const cases=[
['KJ preview VAE/Audio VAE or Tiny VAE is outside the supported configuration',/Keep the generation VAEs connected/],
['Core approximate-VAE preview is outside the supported configuration',/Auto or Latent2RGB/],
];
for(const [index,[reason,expected]] of cases.entries()){
const p=project(1);for(const item of [...p.group_revisions,...p.revisions]){
item.generation_reusable=false;item.resume_disabled_reasons=[reason];
}
const n=e.makeNode(312+index);e.w(n,'generation_mode').value='Review Each Chunk';await e.load(n,p);
assert.match(e.f.reviewStatus(n),expected);
assert.match(e.f.reviewStatus(n),/After correcting the configuration/);
assert(!e.visible(n,'Use it and continue'));
}
});
await test('missing compatibility metadata is not mislabeled v5 or a known KJ error',async e=>{
for(const field of ['sampling_contract_version','generation_reusable']){
const p=project(1);for(const item of [...p.group_revisions,...p.revisions])delete item[field];
const n=e.makeNode(field==='sampling_contract_version'?312:313);e.w(n,'generation_mode').value='Review Each Chunk';await e.load(n,p);
assert.match(e.f.reviewStatus(n),/metadata is incomplete or unsupported/);
assert.doesNotMatch(e.f.reviewStatus(n),/Sampling v5|Bypass KJ/);
assert(!e.visible(n,'Use it and continue'));
}
});
await test('unknown saved reasons do not expose paths or unbounded text',async e=>{
const p=project(1);for(const item of [...p.group_revisions,...p.revisions]){
item.generation_reusable=false;item.resume_disabled_reasons=Array(20).fill('C:/Users/private/file.py'+'x'.repeat(10000));
}
const n=e.makeNode();e.w(n,'generation_mode').value='Review Each Chunk';await e.load(n,p);
const text=e.f.reviewStatus(n);assert.match(text,/Detailed Report/);
assert.doesNotMatch(text,/private|file.py/);assert(text.length<1500);
});
await test('history reasons follow the selected Take rather than the canonical revision',async e=>{
const p=project(2),vae='KJ preview VAE/Audio VAE or Tiny VAE is outside the supported configuration';
const core='Core approximate-VAE preview is outside the supported configuration';
for(const item of [...p.group_revisions,...p.revisions])item.generation_reusable=false;
p.group_revisions[0].resume_disabled_reasons=[vae];p.group_revisions[1].resume_disabled_reasons=[core];
p.revisions[0].resume_disabled_reasons=[core];
const n=e.makeNode();e.w(n,'generation_mode').value='Review Each Chunk';await e.load(n,p);e.f.selectTakeOffset(n,-1);
const text=e.w(n,'Render History / Takes').value;
assert.match(text,/Keep the generation VAEs connected/);assert.doesNotMatch(text,/Auto or Latent2RGB/);
});
await test('saved compatibility metadata controls Review with optional real backend vectors',async e=>{
const configured=process.env.H3_READONLY_PROJECT_VECTORS;
const p=project(1);p.revisions[0].resume_disabled_reasons=['KJ preview runtime code or state could not be verified'];
const vectors=configured?JSON.parse(fs.readFileSync(configured,'utf8')):[{project:p,reusable:true}];
assert(vectors.length>0);
for(const [index,vector] of vectors.entries()){
const n=e.makeNode(800+index,vector.project.run_name);e.w(n,'generation_mode').value='Review Each Chunk';await e.load(n,vector.project);
assert.equal(e.visible(n,'Use it and continue'),vector.reusable,e.f.reviewStatus(n));
if(vector.reason)assert(e.f.reviewStatus(n).includes(vector.reason));
if(vector.reusable)assert.doesNotMatch(e.f.reviewStatus(n),/read-only|could not be verified/);
}
});
// Persisted explicit requests must survive Core's graph-restoration lifecycle.
for(const nodeClass of ['H3ContinuumSamplerV38','H3ContinuumSamplerV39']) {
for(const mode of ['Full Run','Review Each Chunk']) {
for(const boundary of ['Chunk 1','Chunk 2']) {
for(const nonce of [0,5]) {
await test(`regenerate restore: ${nodeClass} ${mode} ${boundary} nonce ${nonce}`,async e=>{
const n=e.makeNode(312,'fixture',nodeClass);await e.load(n,project(3,3));
e.w(n,'generation_mode').value=mode;
e.w(n,'reroll_from_chunk').value=boundary;e.w(n,'reroll_from_chunk').callback?.(boundary);
e.w(n,'reroll_nonce').value=nonce;
const info=JSON.parse(JSON.stringify({properties:n.properties,...n.serialize()}));
const beforeHistory=JSON.stringify(e.getSaved());
n.onRemoved?.();e.app.graph._nodes=[];e.app.extension.beforeConfigureGraph?.();
const loaded=e.makeNode(312,'',nodeClass,false);e.app.extension.nodeCreated(loaded);
loaded.properties=info.properties;loaded.configure(info);
e.app.extension.loadedGraphNode(loaded);await e.finishRestore();
assert.equal(e.w(loaded,'reroll_from_chunk').value,boundary);
assert.equal(e.w(loaded,'reroll_nonce').value,nonce);
await e.queue(loaded);const sent=e.submissions.at(-1).data.output['312'].inputs;
assert.equal(sent.reroll_from_chunk,boundary);assert.equal(sent.reroll_nonce,nonce);
assert.equal(sent.generation_mode,mode);assert.equal(sent.run_storage,'Save + Auto Resume');
assert(!Object.hasOwn(sent,'h3_continuum_regenerate_from'));
assert.equal(JSON.stringify(e.getSaved()),beforeHistory);
});
}
}
}
for(const [boundary,nonce] of [['Chunk 2',0],['Chunk 1',5]]) {
await test(`regenerate legacy: ${nodeClass} ${boundary} nonce ${nonce}`,async e=>{
e.setSaved(project(3,3));e.app.extension.beforeConfigureGraph?.();
const n=e.makeNode(312,'fixture',nodeClass,false);e.app.extension.nodeCreated(n);
e.w(n,'generation_mode').value='Review Each Chunk';e.w(n,'reroll_from_chunk').value=boundary;
e.w(n,'reroll_nonce').value=nonce;e.app.extension.loadedGraphNode(n);await e.finishRestore();
assert.equal(e.w(n,'reroll_from_chunk').value,boundary);assert.equal(e.w(n,'reroll_nonce').value,nonce);
e.w(n,'Run').callback('Review Each Chunk');
await e.queue(n);assert.equal(e.submissions.at(-1).data.output['312'].inputs.reroll_from_chunk,boundary);
assert.equal(e.submissions.at(-1).data.output['312'].inputs.reroll_nonce,nonce);
});
}
await test(`regenerate intent reset: ${nodeClass}`,async e=>{
const n=e.makeNode(312,'fixture',nodeClass);await e.load(n,project(1));
e.w(n,'reroll_from_chunk').value='Chunk 1';e.w(n,'reroll_from_chunk').callback?.('Chunk 1');
assert.equal(n.properties.h3_continuum_regenerate_from,'Chunk 1');
e.w(n,'reroll_from_chunk').value='Auto';e.w(n,'reroll_from_chunk').callback?.('Auto');
assert(!Object.hasOwn(n.properties,'h3_continuum_regenerate_from'));
e.w(n,'Start again from Chunk 1').callback();await e.queue(n);
assert.equal(e.w(n,'reroll_from_chunk').value,'Auto');
assert(!Object.hasOwn(n.properties,'h3_continuum_regenerate_from'));
});
}
console.log(JSON.stringify(results,null,2));if(results.some(r=>!r.pass))process.exitCode=1;
})();
+381
View File
@@ -0,0 +1,381 @@
from __future__ import annotations
import ast
import copy
import json
import subprocess
import sys
import types
from pathlib import Path
import pytest
import torch
from ComfyUI_H3_Continuum_Join import run_storage as storage
from . import test_v31_run_storage as signatures
from . import test_v38_review_run_storage as review
ROOT = Path(__file__).resolve().parents[1]
@pytest.fixture
def core_attention(require_comfy_core, monkeypatch):
import comfy.ldm.modules.attention as attention
import comfy.model_patcher as patcher
# A CPU build may not advertise INT8. Inspect the real loaded definition;
# registration here is a fixture, not an assertion that a CPU kernel runs.
monkeypatch.setitem(attention.REGISTERED_ATTENTION_FUNCTIONS, "comfy_kitchen_int8",
attention.attention_comfy_kitchen_int8)
return attention, patcher
def model_with(core, name):
attention, patcher = core
torch.manual_seed(31)
model = signatures._Model()
model.model_options["transformer_options"] = {}
patcher.ModelPatcher.set_model_optimized_attention(
model, attention.REGISTERED_ATTENTION_FUNCTIONS[name])
return model
def observe(model):
return storage._model_signature(model, "fixture")
def remake(core, monkeypatch, name, *, raw_code=None, defaults=None):
attention, _ = core
old = attention.REGISTERED_ATTENTION_FUNCTIONS[name]
raw = old.__wrapped__
new = types.FunctionType(raw_code or raw.__code__, raw.__globals__,
raw.__name__, raw.__defaults__ if defaults is None else defaults,
raw.__closure__)
new.__module__, new.__qualname__ = raw.__module__, raw.__qualname__
new.__kwdefaults__ = copy.copy(raw.__kwdefaults__)
new.__dict__.update(raw.__dict__)
wrapped = attention.wrap_attn(new)
wrapped.container_function = old.container_function
monkeypatch.setitem(attention.REGISTERED_ATTENTION_FUNCTIONS, name, wrapped)
monkeypatch.setattr(attention, "attention_" + name, wrapped)
return wrapped
@pytest.mark.parametrize("backend", ["pytorch", "comfy_kitchen_int8"])
def test_real_core_backend_is_finite_clone_stable_and_non_mutating(core_attention, monkeypatch, backend):
model = model_with(core_attention, backend)
wrapped = core_attention[0].REGISTERED_ATTENTION_FUNCTIONS[backend]
override = model.model_options["transformer_options"]["optimized_attention_override"]
original = (dict(vars(wrapped)), dict(vars(override)), copy.copy(model.model_options))
watched = {wrapped.__code__, wrapped.__wrapped__.__code__}
if wrapped.container_function:
watched.add(wrapped.container_function.__code__)
calls = []
previous = sys.getprofile()
sys.setprofile(lambda frame, event, arg: calls.append(frame.f_code.co_name)
if event == "call" and frame.f_code in watched else None)
try:
before, safe = observe(model)
finally:
sys.setprofile(previous)
assert safe, before["model_options"]
assert not calls
assert original == (vars(wrapped), vars(override), model.model_options)
cloned, cloned_safe = observe(copy.deepcopy(model))
assert cloned_safe and cloned == before
remake(core_attention, monkeypatch, backend)
recreated, recreated_safe = observe(model_with(core_attention, backend))
assert recreated_safe and recreated == before
attention = before["model_options"]["transformer_options"]["optimized_attention_override"]
assert attention["backend"] == backend
assert attention["core_attention_signature_version"] == 1
assert attention["wrapper"]["closure"][-1]["value"] == {"reference": "wrapper"}
def test_pytorch_and_kitchen_are_distinct(core_attention):
a, safe_a = observe(model_with(core_attention, "pytorch"))
b, safe_b = observe(model_with(core_attention, "comfy_kitchen_int8"))
assert safe_a and safe_b and a != b
@pytest.mark.parametrize("backend", ["pytorch", "comfy_kitchen_int8"])
@pytest.mark.parametrize("change", ["code", "defaults", "helper"])
def test_attention_implementation_changes_identity(core_attention, monkeypatch, backend, change):
before, safe = observe(model_with(core_attention, backend))
assert safe
attention, _ = core_attention
if change == "code":
raw = attention.REGISTERED_ATTENTION_FUNCTIONS[backend].__wrapped__
def change_constant(c):
if isinstance(c, tuple):
return tuple(change_constant(v) for v in c)
return "changed_scale" if c == "scale" else c
constants = tuple(change_constant(c) for c in raw.__code__.co_consts)
assert constants != raw.__code__.co_consts
remake(core_attention, monkeypatch, backend, raw_code=raw.__code__.replace(co_consts=constants))
elif change == "defaults":
raw = attention.REGISTERED_ATTENTION_FUNCTIONS[backend].__wrapped__
remake(core_attention, monkeypatch, backend, defaults=(17,) + raw.__defaults__[1:])
else:
helper = attention._heads_from_dim
constants = tuple(2 if c == 0 else c for c in helper.__code__.co_consts)
monkeypatch.setattr(attention, "_heads_from_dim",
types.FunctionType(helper.__code__.replace(co_consts=constants),
helper.__globals__, helper.__name__))
after, after_safe = observe(model_with(core_attention, backend))
assert after_safe and after != before
@pytest.mark.parametrize("slot", ["inner", "outer"])
@pytest.mark.parametrize("kind", ["function", "opaque", "none"])
def test_container_changes_are_recorded_and_cannot_be_certified(core_attention, monkeypatch, slot, kind):
model = model_with(core_attention, "comfy_kitchen_int8")
before, safe = observe(model)
assert safe
target = (core_attention[0].attention_comfy_kitchen_int8 if slot == "inner" else
model.model_options["transformer_options"]["optimized_attention_override"])
class Opaque:
__slots__ = ()
def __call__(self, *args):
raise AssertionError("signature must not invoke container")
def replacement(*args):
raise AssertionError("signature must not invoke container")
value = {"function": replacement, "opaque": Opaque(), "none": None}[kind]
monkeypatch.setattr(target, "container_function", value)
after, safe_after = observe(model)
assert after != before and not safe_after
def test_unrelated_opaque_cfg_is_not_promoted(core_attention):
model = model_with(core_attention, "pytorch")
assert observe(model)[1]
model.model_options["sampler_cfg_function"] = object()
assert not observe(model)[1]
contract, safe, reasons = signatures._sampling_contract(model=model)
assert contract and not safe and reasons
assert contract["global"]["sampling_contract_version"] == 6
def test_unregistered_or_foreign_setter_keeps_generic_policy(core_attention, monkeypatch):
attention, patcher = core_attention
model = model_with(core_attention, "pytorch")
callback = model.model_options["transformer_options"]["optimized_attention_override"]
monkeypatch.delitem(attention.REGISTERED_ATTENTION_FUNCTIONS, "pytorch")
assert storage._observe_model_options(model.model_options) == storage._observe(model.model_options)
assert not observe(model)[1]
def foreign(_, *args, **kwargs):
return None
model.model_options["transformer_options"]["optimized_attention_override"] = foreign
assert storage._observe_model_options(model.model_options) == storage._observe(model.model_options)
assert model.model_options["transformer_options"]["optimized_attention_override"] is foreign
assert callback is not foreign
def test_missing_provider_and_unknown_native_binding_are_unsafe(core_attention, monkeypatch):
model = model_with(core_attention, "pytorch")
monkeypatch.setattr(torch.nn.functional, "scaled_dot_product_attention", lambda *a: 0)
result, safe = observe(model)
assert not safe
assert "observation_error" in result["model_options"]["transformer_options"]["optimized_attention_override"]
def test_no_override_existing_v6_contract_is_byte_equivalent(monkeypatch):
torch.manual_seed(31)
model, clip, vae = signatures._Model(), signatures._Clip(), signatures._VideoVAE()
model.model_options = {"transformer_options": {}, "disable_cfg1_optimization": True}
# Use the actual pre-repair function, not an expected hash generated by the fix.
original = subprocess.check_output(["git", "-C", str(ROOT), "show", "HEAD:run_storage.py"], text=True)
node = next(n for n in ast.parse(original).body if isinstance(n, ast.FunctionDef) and n.name == "_model_signature")
scope = dict(vars(storage))
exec(compile(ast.Module(body=[node], type_ignores=[]), "original_model_signature.py", "exec"), scope)
before, safe_before, _ = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
with monkeypatch.context() as context:
context.setattr(storage, "_model_signature", scope["_model_signature"])
legacy, safe_legacy, _ = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert safe_before and safe_legacy
assert storage._canonical(before) == storage._canonical(legacy)
assert storage.revision_identity(before) == storage.revision_identity(legacy)
assert storage.SAMPLING_CONTRACT_VERSION == 6 and storage.RUNTIME_SIGNATURE_VERSION == 1
def test_ordinary_custom_override_keeps_existing_descriptor():
model = signatures._Model()
model.model_options = {"transformer_options": {"optimized_attention_override": lambda x: x}}
assert storage._observe_model_options(model.model_options) == storage._observe(model.model_options)
@pytest.mark.parametrize("backend", ["pytorch", "comfy_kitchen_int8"])
def test_same_backend_reuses_prefix_but_changed_backend_does_not(tmp_path, core_attention, backend):
model = model_with(core_attention, backend)
torch.manual_seed(31)
clip, vae = signatures._Clip(), signatures._VideoVAE()
contract, safe, _ = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert safe
source = review._persist(tmp_path, contract, prefix=1, review_unit=(1, 1),
updated_utc="2026-10-14T00:00:00+00:00")
before_files = {str(p):p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
same, same_safe, _ = signatures._sampling_contract(model=copy.deepcopy(model), clip=clip, video_vae=vae, reroll_from_chunk=0)
assert same_safe and storage.revision_identity(same) == storage.revision_identity(contract)
reader = review._controller(tmp_path, same)
reader.configure_review(generation_mode=review.GENERATION_MODE_REVIEW, review_action=review.REVIEW_ACTION_CONTINUE)
prefix = reader._load_validated_review_prefix(same, resume_safe=same_safe)
assert len(prefix.entries) == 1
other = "pytorch" if backend == "comfy_kitchen_int8" else "comfy_kitchen_int8"
changed, changed_safe, _ = signatures._sampling_contract(model=model_with(core_attention, other),
clip=clip, video_vae=vae, reroll_from_chunk=0)
assert changed_safe and storage.revision_identity(changed) != storage.revision_identity(contract)
changed_reader = review._controller(tmp_path, changed)
changed_reader.configure_review(generation_mode=review.GENERATION_MODE_REVIEW, review_action=review.REVIEW_ACTION_CONTINUE)
assert not changed_reader._load_validated_review_prefix(changed, resume_safe=changed_safe).entries
assert before_files == {str(p):p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
@pytest.mark.parametrize("entry_point", ["review", "head", "import", "compatible", "take"])
def test_old_unsafe_attention_manifest_stays_read_only_and_unmodified(tmp_path, core_attention, monkeypatch, entry_point):
model = model_with(core_attention, "pytorch")
options = model.model_options
# Original depth-limited observation, producing the old unsafe v6 contract.
with monkeypatch.context() as context:
context.setattr(storage, "_observe_model_options", storage._observe)
old, old_safe, reasons = signatures._sampling_contract(model=model, reroll_from_chunk=0)
assert not old_safe and reasons
source = review._persist(tmp_path, old, prefix=1, review_unit=(1, 1),
updated_utc="2026-10-14T00:00:00+00:00")
source.manifest["resume_safe"] = False
source.manifest["resume_disabled_reasons"] = reasons
source._write_manifest()
before = {str(p):p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
reader = review._controller(tmp_path, old)
reader.configure_review(generation_mode=review.GENERATION_MODE_REVIEW, review_action=review.REVIEW_ACTION_CONTINUE)
assert storage._reusable_manifest(source.manifest) is False
if entry_point == "review":
assert not reader._load_validated_review_prefix(old, resume_safe=True).entries
elif entry_point == "head":
assert reader.find_latest_review_head(old, resume_safe=True) is None
elif entry_point == "import":
assert reader._find_plan_import(old, resume_safe=True) is None
elif entry_point == "compatible":
assert reader._compatible_plan_prefix(source.manifest, old, resume_safe=True) == ([], [])
else:
reader.take_group = 1
reader.take_revision_id = review._project(source)["canonical_head_revision_id"]
with pytest.raises(storage.RunStorageError, match="read-only"):
reader._validated_take_selection(old, resume_safe=True)
assert before == {str(p):p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
def test_provider_policy_changes_are_observed(core_attention, monkeypatch):
import comfy_kitchen.sage_attention as kitchen_attention
model = model_with(core_attention, "comfy_kitchen_int8")
before, safe = observe(model)
assert safe
monkeypatch.setattr(kitchen_attention, "LARGE_CTA_K", kitchen_attention.LARGE_CTA_K + 1)
after, safe_after = observe(model)
assert safe_after and before != after
@pytest.mark.parametrize("slot", ["setter_globals", "raw_globals", "wrapper_code"])
def test_lookalike_core_structure_is_not_promoted(core_attention, monkeypatch, slot):
attention, _ = core_attention
model = model_with(core_attention, "pytorch")
if slot == "setter_globals":
value = model.model_options["transformer_options"]["optimized_attention_override"]
value = types.FunctionType(value.__code__, dict(value.__globals__), value.__name__,
value.__defaults__, value.__closure__)
model.model_options["transformer_options"]["optimized_attention_override"] = value
elif slot == "raw_globals":
wrapped = attention.attention_pytorch
raw = wrapped.__wrapped__
new = types.FunctionType(raw.__code__, dict(raw.__globals__), raw.__name__,
raw.__defaults__, raw.__closure__)
wrapped = attention.wrap_attn(new)
monkeypatch.setitem(attention.REGISTERED_ATTENTION_FUNCTIONS, "pytorch", wrapped)
monkeypatch.setattr(attention, "attention_pytorch", wrapped)
model = model_with(core_attention, "pytorch")
else:
wrapped = attention.attention_pytorch
monkeypatch.setattr(wrapped, "__code__", wrapped.__code__.replace(
co_consts=wrapped.__code__.co_consts + ("unrecognized wrapper",)))
assert not observe(model)[1]
def test_torch_dispatch_flag_changes_identity(core_attention):
model = model_with(core_attention, "pytorch")
enabled = torch.backends.cuda.math_sdp_enabled()
before, safe = observe(model)
assert safe
try:
torch.backends.cuda.enable_math_sdp(not enabled)
after, after_safe = observe(model)
assert after_safe and after != before
finally:
torch.backends.cuda.enable_math_sdp(enabled)
@pytest.mark.parametrize("backend", ["pytorch", "comfy_kitchen_int8"])
def test_signature_is_stable_in_separate_processes_and_hash_seeds(core_attention, backend):
import os
import comfy
core_root = str(Path(core_attention[1].__file__).resolve().parent.parent)
script = r"""
import importlib.util,json,os,sys
sys.path.insert(0,CORE_ROOT)
os.environ["CUDA_VISIBLE_DEVICES"]="-1"
import torch
torch.cuda._lazy_init=lambda: (_ for _ in ()).throw(RuntimeError("CUDA forbidden"))
from comfy.cli_args import args
args.cpu=True
from comfy.ldm.modules import attention
from comfy import model_patcher
name="ComfyUI_H3_Continuum_Join"
spec=importlib.util.spec_from_file_location(name,ROOT+"/__init__.py",submodule_search_locations=[ROOT])
sys.modules[name]=importlib.util.module_from_spec(spec)
from ComfyUI_H3_Continuum_Join import run_storage as s
from types import SimpleNamespace
class Model:
def __init__(self):
self.model=SimpleNamespace(diffusion_model=torch.nn.Linear(2,2))
self.wrappers={}
self.patches={}
self.model_options={"transformer_options":{}}
def model_dtype(self): return torch.float32
def model_size(self): return 4
torch.manual_seed(31)
model=Model()
attention.REGISTERED_ATTENTION_FUNCTIONS["comfy_kitchen_int8"]=attention.attention_comfy_kitchen_int8
model_patcher.ModelPatcher.set_model_optimized_attention(model,attention.REGISTERED_ATTENTION_FUNCTIONS[BACKEND])
descriptor,safe=s._model_signature(model,"fixture")
assert safe,descriptor["model_options"]
print(s._hash(descriptor))
"""
script = "CORE_ROOT="+repr(core_root)+"\nROOT="+repr(str(ROOT))+"\nBACKEND="+repr(backend)+"\n"+script
results = []
for seed in ("1", "93017"):
result = subprocess.run([sys.executable, "-B", "-c", script],
env={**os.environ, "CUDA_VISIBLE_DEVICES":"-1", "PYTHONHASHSEED":seed},
capture_output=True, text=True, timeout=90)
assert result.returncode == 0, result.stderr + result.stdout
results.append(result.stdout.strip().splitlines()[-1])
assert results[0] == results[1]
def test_unobservable_global_helper_state_stays_unsafe(core_attention, monkeypatch):
attention, _ = core_attention
model = model_with(core_attention, "pytorch")
before, safe = observe(model)
assert safe
scope = vars(attention)
exec("def issue31_helper(*args):\n return ISSUE31_UNOBSERVABLE_STATE", scope)
try:
monkeypatch.setattr(attention, "ISSUE31_UNOBSERVABLE_STATE", object(), raising=False)
monkeypatch.setattr(attention, "_heads_from_dim", scope["issue31_helper"])
after, after_safe = observe(model)
assert after != before and not after_safe
finally:
scope.pop("issue31_helper", None)
def test_fallback_pytorch_implementation_changes_kitchen_identity(core_attention, monkeypatch):
attention, _ = core_attention
model = model_with(core_attention, "comfy_kitchen_int8")
before, safe = observe(model)
assert safe
raw = attention.attention_pytorch.__wrapped__
remake(core_attention, monkeypatch, "pytorch", defaults=(19,) + raw.__defaults__[1:])
after, after_safe = observe(model)
assert after != before
@@ -0,0 +1,328 @@
"""Phase A: real audited KJ wrapper, CPU-only observation and Run Storage gates."""
from __future__ import annotations
import importlib
import json
import os
import sys
import types
from pathlib import Path
import pytest
import torch
from PIL import Image
from ComfyUI_H3_Continuum_Join import preview_override_signature as adapter
from ComfyUI_H3_Continuum_Join import run_storage as storage
from . import test_v31_run_storage as signatures
from . import test_v38_review_run_storage as review
from . import test_issue31_attention_signature as attention
from .test_issue31_attention_signature import core_attention
from .test_issue31_spectrum_signature import spectrum, binding
@pytest.fixture(params=["local", "public"])
def kj(require_comfy_core, request):
root = os.environ.get("H3_KJ_PREVIEW_TEST_ROOT" if request.param == "local" else "H3_KJ_PREVIEW_PUBLIC_TEST_ROOT")
if not root:
pytest.skip("real audited KJ Preview Override provider is not configured")
package = types.ModuleType("issue31_kj_preview_tests_" + request.param)
package.__path__ = [str(Path(root) / "nodes")]
sys.modules[package.__name__] = package
return importlib.import_module(package.__name__ + ".preview_override_node")
def wrapper(kj, **settings):
values = dict(max_resolution=512, node_id="100", jpeg_quality=90,
suppress_default=True, preview_frames=1, preview_fps=12)
values.update(settings)
return kj._PreviewOverrideWrapper(**values)
def model_with(item):
model = signatures._Model()
model.wrappers = {"outer_sample": {"kj_preview_override": [item]}}
return model
def observe(model):
return storage._model_signature(model, "fixture")
def test_no_kj_observer_and_full_sampling_contract_are_exact(monkeypatch):
model = signatures._Model()
model.wrappers = {"outer_sample": {"other": [lambda executor: executor()]}}
current = observe(model)
torch.manual_seed(31016)
clip, vae = signatures._Clip(), signatures._VideoVAE()
contract = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
with monkeypatch.context() as ctx:
ctx.setattr(storage, "observe_model_wrappers", lambda value, observer: observer(value))
assert observe(model) == current
old = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert storage._canonical(contract) == storage._canonical(old)
assert storage.revision_identity(contract[0]) == storage.revision_identity(old[0])
def test_real_provider_frames_and_recreated_wrapper_keep_contract(kj):
item = wrapper(kj)
model = model_with(item)
before = observe(model)
assert before[1], before[0]
clip, vae = signatures._Clip(), signatures._VideoVAE()
original = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert original[1]
for count in (1, 3, 0):
item.frames[:] = [Image.new("RGB", (2, 2)) for _ in range(count)]
frames = item.frames
state = dict(vars(item))
assert observe(model) == before
current = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert current == original
assert vars(item) == state and item.frames is frames
model.wrappers["outer_sample"]["kj_preview_override"] = [wrapper(kj)]
assert observe(model) == before
@pytest.mark.parametrize("setting,value", [("max_resolution", 768), ("node_id", "101"),
("jpeg_quality", 80), ("suppress_default", False), ("preview_frames", 4), ("preview_fps", 24)])
def test_every_fixed_setting_changes_identity(kj, setting, value):
item = wrapper(kj)
before, safe = adapter.preview_override_signature(item)
assert safe, before
setattr(item, setting, value)
after, safe = adapter.preview_override_signature(item)
assert safe and after != before
def test_full_audit_profile_is_part_of_identity(kj, monkeypatch):
item = wrapper(kj)
before, safe = adapter.preview_override_signature(item)
assert safe
# An audit revision must change identity even with the same display ID and
# wrapper source, so dependency-profile upgrades cannot reuse old Takes.
profile, _ = adapter._provider_source(kj.__file__)
monkeypatch.setitem(profile, "audit_revision", 2)
after, safe = adapter.preview_override_signature(item)
assert safe and before["profile"] == after["profile"]
assert before["profile_sha256"] != after["profile_sha256"]
def other_wrapper(executor, *args, **kwargs):
return executor(*args, **kwargs)
def test_key_and_list_execution_order_change_identity(kj):
a, b = wrapper(kj), wrapper(kj, jpeg_quality=80)
model = model_with(a)
model.wrappers["outer_sample"] = {"other": [other_wrapper], "kj_preview_override": [a, b]}
before, safe = observe(model)
assert safe
model.wrappers["outer_sample"] = {"kj_preview_override": [a, b], "other": [other_wrapper]}
after, safe = observe(model)
assert safe and after != before
model.wrappers["outer_sample"]["kj_preview_override"] = [b, a]
reordered, safe = observe(model)
assert safe and reordered != after
@pytest.mark.parametrize("change", ["source", "call_code", "helper_code", "defaults", "class_attribute",
"instance_attribute", "frames_tuple", "frames_subclass", "vae", "audio_vae", "tiny_vae",
"foreign", "subclass", "invalid_setting", "core_factory", "imported_helper_code",
"core_preview_code", "dispatcher_code", "core_tiny_vae"])
def test_unknown_implementations_and_out_of_scope_state_reject_reuse(kj, monkeypatch, change):
item = wrapper(kj)
if change == "source":
profile, _ = adapter._provider_source(kj.__file__)
monkeypatch.setitem(profile, "source_sha256", "f" * 64)
elif change in ("call_code", "helper_code"):
fn = type(item).__call__ if change == "call_code" else kj._normalize_packed_x0
monkeypatch.setattr(fn, "__code__", fn.__code__.replace(co_consts=fn.__code__.co_consts + ("changed",)))
elif change == "defaults":
monkeypatch.setattr(type(item).__init__, "__defaults__", (2, 12, None, "none", None))
elif change == "class_attribute":
monkeypatch.setattr(type(item), "unrecognized", True, raising=False)
elif change == "instance_attribute":
item.unrecognized = True
elif change == "frames_tuple":
item.frames = ()
elif change == "frames_subclass":
class Frames(list):
pass
item.frames = Frames()
elif change in ("vae", "audio_vae"):
setattr(item, change, object())
elif change == "tiny_vae":
item.tiny_vae = "taeh3"
elif change == "foreign":
item = types.SimpleNamespace(**vars(item))
elif change == "subclass":
class Extra(type(item)):
pass
item = Extra(512, "100", 90, True)
elif change == "invalid_setting":
item.preview_frames = True
elif change == "core_factory":
monkeypatch.setattr(kj.latent_preview, "get_previewer", lambda *a: None)
elif change == "core_tiny_vae":
import latent_preview
monkeypatch.setattr(latent_preview.args, "preview_method", latent_preview.LatentPreviewMethod.TAESD)
else:
if change == "imported_helper_code":
fn = kj.load_tiny_vae_decoder
elif change == "core_preview_code":
fn = kj.latent_preview.Latent2RGBPreviewer.decode_latent_to_preview
else:
fn = sys.modules["comfy.patcher_extension"].get_all_wrappers
monkeypatch.setattr(fn, "__code__", fn.__code__.replace(co_consts=fn.__code__.co_consts + ("changed",)))
assert not observe(model_with(item))[1]
def test_unknown_other_wrapper_is_not_promoted(kj):
model = model_with(wrapper(kj))
model.wrappers["outer_sample"]["unknown"] = [torch.nn.Linear(2, 2)]
assert not observe(model)[1]
@pytest.mark.parametrize("items", [None, (), [], object()])
def test_unknown_target_container_is_not_promoted(items):
assert not adapter.observe_model_wrappers({"outer_sample": {"kj_preview_override": items}}, storage._observe)[1]
@pytest.mark.parametrize("container", ["root", "outer", "items"])
def test_observation_limit_rejects_reuse_without_interrupting_generation(container):
wrappers = {"outer_sample": {"kj_preview_override": [object()]}}
if container == "items":
wrappers["outer_sample"]["kj_preview_override"] *= 4097
else:
target = wrappers if container == "root" else wrappers["outer_sample"]
target.update({str(i): [] for i in range(4097)})
result, safe = adapter.observe_model_wrappers(wrappers, storage._observe)
assert not safe and result["unsupported"]
def test_observation_never_executes_provider_or_initializes_cuda(kj, monkeypatch):
item = wrapper(kj)
# Even frame objects whose representation or data cannot be inspected are
# ignored. The exact builtin list is the only permitted transient container.
class UnreadableFrame:
def __repr__(self):
raise AssertionError("frame inspected")
item.frames.append(UnreadableFrame())
def forbidden(*args, **kwargs):
raise AssertionError("signature tried to execute GPU/preview code")
monkeypatch.setattr(torch.cuda, "_lazy_init", forbidden)
# Trace calls instead of patching the audited callable (which must reject).
called = []
watched = {kj._PreviewOverrideWrapper.__call__.__code__, kj._get_core_previewer.__code__,
kj._frame_to_pil.__code__, kj.load_tiny_vae_decoder.__code__}
def trace(frame, event, arg):
if event == "call" and frame.f_code in watched:
called.append(frame.f_code.co_name)
return trace
previous = sys.gettrace()
try:
sys.settrace(trace)
result, safe = observe(model_with(item))
finally:
sys.settrace(previous)
assert safe, result
assert called == [] and not torch.cuda.is_initialized()
@pytest.mark.parametrize("backend", ["pytorch", "comfy_kitchen_int8"])
def test_attention_and_spectrum_combination_is_stable(kj, core_attention, spectrum, backend):
model = attention.model_with(core_attention, backend)
model.model_options["spectrum_h3_binding"] = binding(spectrum)
base, safe = observe(model)
assert safe
item = wrapper(kj)
model.wrappers = {"outer_sample": {"kj_preview_override": [item]}}
before, safe = observe(model)
assert safe
assert before["model_options"] == base["model_options"]
item.frames.append(Image.new("RGB", (2, 2)))
assert observe(model) == (before, safe)
def test_saved_prefix_is_reusable_after_preview_only_change(kj, tmp_path):
item = wrapper(kj)
model = model_with(item)
clip, vae = signatures._Clip(), signatures._VideoVAE()
contract, safe, _ = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert safe
saved = signatures._persist_contract_prefix(tmp_path, contract, count=1)
before = {str(p): p.read_bytes() for p in tmp_path.rglob("*") if p.is_file()}
item.frames.append(Image.new("RGB", (2, 2)))
current, safe, _ = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert safe and storage.revision_identity(current) == storage.revision_identity(contract)
_, entries, records = signatures._load_compatible_prefix(tmp_path, saved, current)
assert len(entries) == len(records) == 1
assert before == {str(p): p.read_bytes() for p in tmp_path.rglob("*") if p.is_file()}
@pytest.mark.parametrize("entry_point", ["review", "head", "import", "compatible", "take"])
def test_old_unsafe_preview_history_is_never_promoted(kj, tmp_path, monkeypatch, entry_point):
item = wrapper(kj)
item.frames.append(Image.new("RGB", (2, 2)))
model = model_with(item)
with monkeypatch.context() as ctx:
ctx.setattr(storage, "observe_model_wrappers", lambda value, observer: observer(value))
old, safe, reasons = signatures._sampling_contract(model=model, reroll_from_chunk=0)
assert not safe
source = review._persist(tmp_path, old, prefix=1, review_unit=(1, 1), updated_utc="2026-10-16T00:00:00+00:00")
source.manifest["resume_safe"] = False
source.manifest["resume_disabled_reasons"] = reasons
source._write_manifest()
before = {str(p): p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
reader = review._controller(tmp_path, old)
reader.configure_review(generation_mode=review.GENERATION_MODE_REVIEW, review_action=review.REVIEW_ACTION_CONTINUE)
assert not storage._reusable_manifest(source.manifest)
if entry_point == "review": assert not reader._load_validated_review_prefix(old, resume_safe=True).entries
elif entry_point == "head": assert reader.find_latest_review_head(old, resume_safe=True) is None
elif entry_point == "import": assert reader._find_plan_import(old, resume_safe=True) is None
elif entry_point == "compatible": assert reader._compatible_plan_prefix(source.manifest, old, resume_safe=True) == ([], [])
else:
reader.take_group = 1
reader.take_revision_id = review._project(source)["canonical_head_revision_id"]
with pytest.raises(storage.RunStorageError, match="read-only"):
reader._validated_take_selection(old, resume_safe=True)
assert before == {str(p): p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
def test_other_profile_registration_does_not_change_identity(kj, monkeypatch):
item = wrapper(kj)
before, safe = adapter.preview_override_signature(item)
assert safe
active, _ = adapter._provider_source(kj.__file__)
monkeypatch.setattr(adapter, "_PROFILES", (active,))
assert adapter.preview_override_signature(item) == (before, True)
monkeypatch.setattr(adapter, "_PROFILES", (active, dict(active, source_sha256="e" * 64)))
assert adapter.preview_override_signature(item) == (before, True)
monkeypatch.setattr(adapter, "_PROFILES", (active, active))
assert not adapter.preview_override_signature(item)[1]
@pytest.mark.parametrize("contents", [None, "invalid json", "null", "[]", "{}"])
def test_missing_or_malformed_optional_profile_does_not_break_existing_paths(tmp_path, contents):
path = tmp_path / "public.json"
if contents is not None:
path.write_text(contents, encoding="utf-8")
assert adapter._load_public_profile(path) is None
model = signatures._Model()
assert observe(model)[1]
@pytest.mark.parametrize("field", ["id", "source_sha256"])
def test_optional_profile_collision_cannot_disable_original_profile(kj, tmp_path, monkeypatch, field):
candidate = dict(adapter._PUBLIC_PROFILE)
candidate[field] = adapter._PROFILE[field]
path = tmp_path / "public.json"
path.write_text(json.dumps(candidate), encoding="utf-8")
loaded = adapter._load_public_profile(path)
assert loaded is None
monkeypatch.setattr(adapter, "_PROFILES", (adapter._PROFILE,))
item = wrapper(kj)
expected = "_local." in kj.__name__
assert adapter.preview_override_signature(item)[1] is expected
assert observe(signatures._Model())[1]
+329
View File
@@ -0,0 +1,329 @@
from __future__ import annotations
import ast
import dataclasses
import hashlib
import importlib
import json
import os
import subprocess
import sys
import types
from pathlib import Path
import pytest
import torch
from ComfyUI_H3_Continuum_Join import run_storage as storage
from ComfyUI_H3_Continuum_Join import spectrum_signature as adapter
from . import test_issue31_attention_signature as attention_tests
from .test_issue31_attention_signature import core_attention
from . import test_v31_run_storage as signatures
from . import test_v38_review_run_storage as review
@pytest.fixture(params=["local", "public029"])
def spectrum(require_comfy_core, request):
root = os.environ.get("H3_SPECTRUM_TEST_ROOT" if request.param == "local" else "H3_SPECTRUM_PUBLIC_TEST_ROOT")
if not root:
pytest.skip("real audited Spectrum checkout is not configured")
package = types.ModuleType("issue31_spectrum_tests_" + request.param)
package.__path__ = [root]
sys.modules[package.__name__] = package
prefix = package.__name__ + ".comfyui_spectrum_h3"
sampling = importlib.import_module(prefix + ".sampling")
runtime = importlib.import_module(prefix + ".runtime")
config = importlib.import_module(prefix + ".config")
return sampling, runtime, config
def binding(spectrum, **settings):
sampling, runtime, config = spectrum
return sampling.SpectrumH3Binding(runtime.SpectrumH3Runtime(config.SpectrumH3Config(**settings)))
def signature(value):
result, safe = adapter.spectrum_binding_signature(value)
assert safe, result.get("unsupported")
return result
def test_recreated_runtime_and_transient_history_keep_identity(spectrum):
a, b = binding(spectrum), binding(spectrum)
before = signature(a)
assert signature(b) == before
a.runtime._run_counter += 8
a.runtime._consecutive_forecasts = 3
a.runtime._current_window = 3.25
a.runtime.forecaster.update(0.0, torch.arange(12, dtype=torch.float32).reshape(1,12))
a.runtime.forecaster.update(1.0, torch.arange(12, dtype=torch.float32).reshape(1,12) + 1)
a.runtime.model_aware.anchor_count = 7
a.runtime.stats.current_window = 4.0
assert signature(a) == before
CHANGES = {
"enabled": {"enabled": False}, "blend_weight": {"blend_weight": 0.6},
"degree": {"degree": 2, "bootstrap_first_forecast": False},
"ridge_lambda": {"ridge_lambda": 0.2}, "window_size": {"window_size": 3.0},
"flex_window": {"flex_window": 1.0},
"warmup_steps": {"warmup_steps": 2, "bootstrap_first_forecast": False},
"tail_actual_steps": {"tail_actual_steps": 2}, "max_history": {"max_history": 9},
"history_storage": {"history_storage": "vram"}, "debug": {"debug": True},
"force_actual": {"force_actual": True},
"bootstrap_first_forecast": {"bootstrap_first_forecast": False},
"anchor_residual_feedback": {"anchor_residual_feedback": True, "offline_smoothing_replay": False},
"selective_rollback_correction": {"selective_rollback_correction": True, "offline_smoothing_replay": False},
"offline_smoothing_replay": {"offline_smoothing_replay": False},
"audio_blend_weight": {"audio_blend_weight": 0.1},
"offline_archive_storage": {"offline_archive_storage": "vram"},
"model_aware_mode": {"model_aware_mode": "full"},
"model_aware_risk_threshold": {"model_aware_risk_threshold": 0.5},
"model_aware_trust_shrinkage": {"model_aware_mode": "full", "model_aware_trust_shrinkage": True},
"model_aware_replay_generic_correction": {"model_aware_replay_generic_correction": True},
"generic_correction_mode": {"generic_correction_mode": "legacy"},
"generic_correction_limiter": {"generic_correction_limiter": "tanh"},
"generic_correction_limit": {"generic_correction_limit": 0.3},
"generic_correction_attenuation": {"generic_correction_attenuation": "general_confidence"},
"sa_pece_forecast_policy": {"sa_pece_forecast_policy": "stable_start"},
}
@pytest.mark.parametrize("field", sorted(CHANGES))
def test_every_config_field_changes_identity(spectrum, field):
before = signature(binding(spectrum))
changed = signature(binding(spectrum, **CHANGES[field]))
assert set(before["config"]) == set(CHANGES) == set(adapter.CONFIG_FIELDS)
assert before != changed and before["config"][field] != changed["config"][field]
@pytest.mark.parametrize("change", ["runtime_code", "wrapper_code", "wrapper_closure", "class_attribute", "source_digest"])
def test_implementation_changes_are_rejected(spectrum, monkeypatch, change):
item = binding(spectrum)
signature(item)
if change == "runtime_code":
old = spectrum[1].SpectrumH3Runtime.__init__
monkeypatch.setattr(old, "__code__", old.__code__.replace(co_consts=old.__code__.co_consts + ("changed",)))
elif change == "wrapper_code":
old = spectrum[0].outer_sample_wrapper
monkeypatch.setattr(old, "__code__", old.__code__.replace(co_consts=old.__code__.co_consts + ("changed",)))
elif change == "wrapper_closure":
old = spectrum[0].run_selective_rollback_euler
assert old.__closure__
cells = list(old.__closure__)
cells[0] = (lambda value: lambda: value)(object()).__closure__[0]
changed = types.FunctionType(old.__code__, old.__globals__, old.__name__, old.__defaults__, tuple(cells))
changed.__module__, changed.__qualname__ = old.__module__, old.__qualname__
monkeypatch.setattr(spectrum[0], "run_selective_rollback_euler", changed)
elif change == "class_attribute":
monkeypatch.setattr(spectrum[1].SpectrumH3Runtime, "extra_contract", 1, raising=False)
else:
source, _ = adapter._select_source_profile(Path(spectrum[0].__file__).resolve().parent)
monkeypatch.setitem(source, "sampling.py", "f" * 64)
after, safe = adapter.spectrum_binding_signature(item)
assert not safe and after.get("unsupported")
@pytest.mark.parametrize("change", ["foreign_binding", "binding_subclass", "runtime_extra", "component_extra", "component_setting", "nan_config", "stats_replaced", "missing_runtime_field"])
def test_unknown_binding_and_additional_state_are_not_promoted(spectrum, change):
item = binding(spectrum)
if change == "foreign_binding":
@dataclasses.dataclass(slots=True)
class SpectrumH3Binding:
runtime: object
item = SpectrumH3Binding(item.runtime)
elif change == "binding_subclass":
class ExtraBinding(type(item)):
pass
item = ExtraBinding(item.runtime)
elif change == "runtime_extra":
item.runtime.unrecognized = object()
elif change == "component_extra":
item.runtime.forecaster.unrecognized = 1
elif change == "component_setting":
item.runtime.forecaster.degree += 1
elif change == "stats_replaced":
item.runtime.stats = object()
elif change == "missing_runtime_field":
del item.runtime._run_counter
else:
object.__setattr__(item.runtime.config, "blend_weight", float("nan"))
assert adapter.spectrum_binding_signature(item)[1] is False
@pytest.mark.parametrize("backend", ["pytorch", "comfy_kitchen_int8"])
def test_attention_and_spectrum_remain_clone_stable(core_attention, spectrum, backend):
model = attention_tests.model_with(core_attention, backend)
original = attention_tests.observe(model)
model.model_options["spectrum_h3_binding"] = binding(spectrum)
combined = attention_tests.observe(model)
assert original[1] and combined[1]
assert combined[0]["model_options"]["transformer_options"] == original[0]["model_options"]["transformer_options"]
model.model_options["spectrum_h3_binding"] = binding(spectrum)
assert attention_tests.observe(model) == combined
@pytest.mark.parametrize("backend", [None, "pytorch", "comfy_kitchen_int8"])
def test_spectrum_absent_preserves_full_phase1_descriptor(core_attention, monkeypatch, backend):
model = signatures._Model() if backend is None else attention_tests.model_with(core_attention, backend)
current = attention_tests.observe(model)
old = storage._observe_attention_model_options(model.model_options)
assert storage._observe_model_options(model.model_options) == old
# Independent Phase 1 implementation is frozen in the reviewed test fixture.
source = (Path(__file__).parent / "fixtures" / "issue31_phase1_model_options.py.txt").read_text()
scope = dict(vars(storage))
exec(compile(source, "phase1_options_fixture", "exec"), scope)
before = scope["_observe_model_options"](model.model_options)
assert before == old
torch.manual_seed(315)
clip, vae = signatures._Clip(), signatures._VideoVAE()
contract, safe, reasons = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
with monkeypatch.context() as context:
context.setattr(storage, "_observe_model_options", scope["_observe_model_options"])
previous, previous_safe, previous_reasons = signatures._sampling_contract(model=model, clip=clip, video_vae=vae, reroll_from_chunk=0)
assert safe == previous_safe and reasons == previous_reasons
assert storage._canonical(contract) == storage._canonical(previous)
assert storage.revision_identity(contract) == storage.revision_identity(previous)
@pytest.mark.parametrize("entry_point", ["review", "head", "import", "compatible", "take"])
def test_old_unsafe_spectrum_history_is_unmodified(tmp_path, core_attention, spectrum, monkeypatch, entry_point):
model = attention_tests.model_with(core_attention, "pytorch")
model.model_options["spectrum_h3_binding"] = binding(spectrum)
with monkeypatch.context() as ctx:
ctx.setattr(storage, "_observe_model_options", storage._observe_attention_model_options)
old, safe, reasons = signatures._sampling_contract(model=model, reroll_from_chunk=0)
assert not safe and reasons
source = review._persist(tmp_path, old, prefix=1, review_unit=(1,1), updated_utc="2026-10-15T00:00:00+00:00")
source.manifest["resume_safe"] = False
source.manifest["resume_disabled_reasons"] = reasons
source._write_manifest()
before = {str(p):p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
current, current_safe, _ = signatures._sampling_contract(model=model, reroll_from_chunk=0)
assert current_safe and storage.revision_identity(current) != storage.revision_identity(old)
reader = review._controller(tmp_path, old)
reader.configure_review(generation_mode=review.GENERATION_MODE_REVIEW, review_action=review.REVIEW_ACTION_CONTINUE)
assert not storage._reusable_manifest(source.manifest)
if entry_point == "review": assert not reader._load_validated_review_prefix(old,resume_safe=True).entries
elif entry_point == "head": assert reader.find_latest_review_head(old,resume_safe=True) is None
elif entry_point == "import": assert reader._find_plan_import(old,resume_safe=True) is None
elif entry_point == "compatible": assert reader._compatible_plan_prefix(source.manifest,old,resume_safe=True)==([],[])
else:
reader.take_group=1;reader.take_revision_id=review._project(source)["canonical_head_revision_id"]
with pytest.raises(storage.RunStorageError,match="read-only"):reader._validated_take_selection(old,resume_safe=True)
assert before=={str(p):p.read_bytes() for p in source.run_root.rglob("*") if p.is_file()}
def test_imports_and_profile_cache_do_not_change_identity(spectrum, monkeypatch):
item=binding(spectrum);before=signature(item)
prefix=spectrum[0].__name__.rsplit('.',1)[0]
importlib.import_module(prefix+'.experiments')
aware=importlib.import_module(prefix+'.model_aware')
monkeypatch.setitem(aware._PROFILE_CACHE,('temporary',),object())
assert signature(item)==before
def test_imported_original_implementation_changes_are_rejected(spectrum, monkeypatch):
item=binding(spectrum);signature(item)
module=importlib.import_module(spectrum[0].__name__.rsplit('.',1)[0]+'.generic_correction')
old=module._ORIGINAL_FORECASTER_UPDATE
changed=types.FunctionType(old.__code__.replace(co_consts=old.__code__.co_consts+('changed imported original',)),old.__globals__,old.__name__,old.__defaults__,old.__closure__)
changed.__module__,changed.__qualname__=old.__module__,old.__qualname__
monkeypatch.setattr(module,'_ORIGINAL_FORECASTER_UPDATE',changed)
assert adapter.spectrum_binding_signature(item)[1] is False
@pytest.mark.parametrize('backend',['pytorch','comfy_kitchen_int8'])
def test_spectrum_same_settings_reuse_and_changed_settings_reject(tmp_path,core_attention,spectrum,backend):
model=attention_tests.model_with(core_attention,backend)
model.model_options['spectrum_h3_binding']=binding(spectrum)
torch.manual_seed(315);clip,vae=signatures._Clip(),signatures._VideoVAE()
contract,safe,_=signatures._sampling_contract(model=model,clip=clip,video_vae=vae,reroll_from_chunk=0);assert safe
source=review._persist(tmp_path,contract,prefix=1,review_unit=(1,1),updated_utc='2026-10-15T00:00:00+00:00')
before={str(p):p.read_bytes() for p in source.run_root.rglob('*') if p.is_file()}
reader=review._controller(tmp_path,contract);reader.configure_review(generation_mode=review.GENERATION_MODE_REVIEW,review_action=review.REVIEW_ACTION_CONTINUE)
assert len(reader._load_validated_review_prefix(contract,resume_safe=True).entries)==1
model.model_options['spectrum_h3_binding']=binding(spectrum,blend_weight=0.6)
changed,changed_safe,_=signatures._sampling_contract(model=model,clip=clip,video_vae=vae,reroll_from_chunk=0);assert changed_safe
other=review._controller(tmp_path,changed);other.configure_review(generation_mode=review.GENERATION_MODE_REVIEW,review_action=review.REVIEW_ACTION_CONTINUE)
assert not other._load_validated_review_prefix(changed,resume_safe=True).entries
assert before=={str(p):p.read_bytes() for p in source.run_root.rglob('*') if p.is_file()}
def test_signature_stable_across_processes_and_module_aliases(spectrum,core_attention):
root=str(Path(spectrum[0].__file__).resolve().parents[1])
core_root=str(Path(core_attention[1].__file__).resolve().parents[1])
adapter_path=str(Path(adapter.__file__).resolve())
script=r"""
import os,sys,types,importlib,importlib.util,json,hashlib
os.environ['CUDA_VISIBLE_DEVICES']='-1'
sys.path.insert(0,CORE_ROOT)
import torch
torch.cuda._lazy_init=lambda: (_ for _ in ()).throw(RuntimeError('CUDA forbidden'))
from comfy.cli_args import args
args.cpu=True
package=types.ModuleType('process_alias_'+os.environ['PYTHONHASHSEED']);package.__path__=[PROVIDER_ROOT];sys.modules[package.__name__]=package
prefix=package.__name__+'.comfyui_spectrum_h3'
s=importlib.import_module(prefix+'.sampling');r=importlib.import_module(prefix+'.runtime');c=importlib.import_module(prefix+'.config')
spec=importlib.util.spec_from_file_location('process_adapter',ADAPTER_FILE);a=importlib.util.module_from_spec(spec);spec.loader.exec_module(a)
value,safe=a.spectrum_binding_signature(s.SpectrumH3Binding(r.SpectrumH3Runtime(c.SpectrumH3Config())))
assert safe,value.get('unsupported')
print(hashlib.sha256(json.dumps(value,sort_keys=True,separators=(',',':')).encode()).hexdigest())
"""
script=script.replace('CORE_ROOT',repr(core_root)).replace('PROVIDER_ROOT',repr(root)).replace('ADAPTER_FILE',repr(adapter_path))
outputs=[]
for seed in ('13','71'):
env=dict(os.environ,PYTHONHASHSEED=seed,PYTHONDONTWRITEBYTECODE='1')
output=subprocess.check_output([sys.executable,'-c',script],env=env,text=True)
outputs.append(output.strip().splitlines()[-1])
assert outputs[0]==outputs[1]
def test_core_node_registration_location_is_not_generation_identity(spectrum,monkeypatch):
item=binding(spectrum);before=signature(item)
prefix=spectrum[0].__name__.rsplit('.',1)[0]
for suffix in ['nodes','objective_media_nodes']:
module=importlib.import_module(prefix+'.'+suffix)
for cls in module.NODE_CLASS_MAPPINGS.values():
monkeypatch.setattr(cls,'RELATIVE_PYTHON_MODULE','custom_nodes.test_location',raising=False)
assert signature(item)==before
def test_original_local_profile_descriptor_is_unchanged(spectrum):
item=binding(spectrum)
source, _ = adapter._select_source_profile(Path(spectrum[0].__file__).resolve().parent)
if source is not adapter.SUPPORTED_SOURCE:
pytest.skip("original descriptor fixture covers the pre-existing local profile")
scope={"__file__":adapter.__file__}
fixture=Path(__file__).parent/"fixtures"/"issue31_local_spectrum_signature.py.txt"
exec(compile(fixture.read_text(),str(fixture),"exec"),scope)
previous = scope["spectrum_binding_signature"](item)
current = adapter.spectrum_binding_signature(item)
if sys.version_info[:2] == (3, 13):
assert previous == current
else:
# Python 3.12 was never accepted by the original 3.13-only profile.
# New support changes current observation, never a stored unsafe Take.
assert sys.version_info[:2] == (3, 12)
assert previous[1] is False and current[1] is True
def test_foreign_provider_file_set_stays_unsafe(tmp_path):
(tmp_path/"unknown.py").write_text("pass")
with pytest.raises(ValueError,match="unsupported provider source set"):
adapter._select_source_profile(tmp_path)
def test_foreign_module_closure_is_rejected():
with pytest.raises(ValueError, match="unsupported provider module closure"):
adapter._callable_value(types.ModuleType("foreign_provider"),provider_prefix="known.comfyui_spectrum_h3")
def test_unloaded_module_cannot_impersonate_provider():
with pytest.raises(ValueError, match="provider module closure is not loaded"):
adapter._callable_value(types.ModuleType("known.comfyui_spectrum_h3.runtime"),provider_prefix="known.comfyui_spectrum_h3")
def test_loaded_implementation_profiles_are_packaged():
root=Path(adapter.__file__).parent
for filename in ["spectrum_provider_v1.json","spectrum_provider_v029.json"]:
assert json.loads((root/filename).read_text())["format"]=="audited_spectrum_loaded_implementation_v1"
assert len(adapter.SUPPORTED_SOURCE)==44 and len(adapter.PUBLIC_SOURCE_029)==55
+162
View File
@@ -0,0 +1,162 @@
"""Saved reason projection and real-provider Review continuation, CPU only."""
import copy
import hashlib
import json
from pathlib import Path
import pytest
import torch
from ComfyUI_H3_Continuum_Join import run_storage as storage
from ComfyUI_H3_Continuum_Join import preview_override_signature as adapter
from . import test_v31_run_storage as signatures
from . import test_v38_review_run_storage as review
from .test_issue31_preview_override_signature import kj, wrapper, model_with
def prepare_kwargs(model):
return dict(model=model, model_fingerprint_value="f" * 64, clip=signatures._Clip(),
video_vae=signatures._VideoVAE(), sampler=signatures._Sampler(), sigmas=torch.tensor([1., 0.]),
prompt_plan={"chunks": 3, "hashes": [str(i) * 64 for i in (1, 2, 3)],
"mode": "list", "prompts": ["prompt 1", "prompt 2", "prompt 3"]},
width=96, height=64, chunk_seconds=5., continuity="Balanced", audio_continuity=True,
base_seed=1, reroll_from_chunk=0, reroll_nonce=0, first_frame_hash="a" * 64,
last_frame_hash="", identity_hash="fixture", strict_compatibility=True,
existing_session=None, conditioning_mode="i2va")
def controller_at(path):
controller = storage.RunStorageController("preview-reuse-test")
controller.run_root = path
controller.revisions_root = path / "revisions"
controller.set_prompt_graph(signatures._prompt_graph(), "42")
controller.configure_review(generation_mode=review.GENERATION_MODE_REVIEW,
review_action=review.REVIEW_ACTION_CONTINUE)
return controller
@pytest.mark.parametrize("setting", [None, "vae", "audio_vae", "tiny_vae"])
def test_real_provider_saved_take_and_separate_controller_continue(kj, tmp_path, setting):
item = wrapper(kj)
if setting is not None:
setattr(item, setting, "taeh3" if setting == "tiny_vae" else object())
kwargs = prepare_kwargs(model_with(item))
source = controller_at(tmp_path)
source.prepare(**kwargs)
entry = review._entry(0, source.contract)
source.commit_group((entry,), positions=(0,))
source.mark_review_group(start=1, end=1, physical_group=1)
source.finalize(session={"session_id": "fixture", "chunks": [entry]}, report="CPU fixture",
review_pause_metadata=source.review_pause_metadata())
project = review._project(source)
take = project["group_revisions"][0]
expected_safe = setting is None
assert take["generation_reusable"] is expected_safe
assert take["sampling_contract_version"] == 6
if expected_safe:
assert "resume_disabled_reasons" not in take
else:
expected = "KJ preview VAE/Audio VAE or Tiny VAE is outside the supported configuration"
assert expected in source.manifest["resume_disabled_reasons"]
assert expected in take["resume_disabled_reasons"]
assert take["resume_disabled_reasons"] == project["revisions"][0]["resume_disabled_reasons"]
original_raw = {str(p): hashlib.sha256(p.read_bytes()).hexdigest() for p in tmp_path.rglob("*.safetensors")}
original_identity = storage.revision_identity(source.contract)
item.frames.append(object())
again = controller_at(tmp_path)
session = again.prepare(**kwargs)
assert again.reused_count == int(expected_safe)
assert len((session or {}).get("chunks", [])) == int(expected_safe)
assert again.review_execution.next_review_unit_start == (2 if expected_safe else 1)
if expected_safe:
assert storage.revision_identity(again.contract) == original_identity
assert original_raw and all(hashlib.sha256(Path(p).read_bytes()).hexdigest() == digest
for p, digest in original_raw.items())
@pytest.mark.parametrize("failure,expected", [
("provider source is outside the audited profile", "KJ/Core preview implementation is outside the audited source profiles"),
("Core approximate VAE is outside Phase A", "Core approximate-VAE preview is outside the supported configuration"),
("loaded code differs from audited source", "KJ preview runtime code or state could not be verified"),
("private /local/path exception", "KJ preview runtime code or state could not be verified"),
])
def test_reason_classification_uses_only_observed_slot(failure, expected):
descriptor = {"wrappers": {"outer_sample": {"kj_preview_override": [
{"adapter": "kj_preview_override_v1", "unsupported": failure}]}}}
before = copy.deepcopy(descriptor)
assert adapter.preview_override_reuse_reasons(descriptor) == [expected]
assert descriptor == before
assert adapter.preview_override_reuse_reasons({"other": descriptor}) == []
def test_reasons_do_not_enter_sampling_identity(kj, monkeypatch):
item = wrapper(kj)
item.tiny_vae = "taeh3"
model, clip, vae = model_with(item), signatures._Clip(), signatures._VideoVAE()
original, safe, reasons = signatures._sampling_contract(model=model, clip=clip, video_vae=vae)
assert not safe and any(reason in adapter.PREVIEW_REUSE_REASONS for reason in reasons)
monkeypatch.setattr(storage, "preview_override_reuse_reasons", lambda descriptor: [])
current, safe, generic = signatures._sampling_contract(model=model, clip=clip, video_vae=vae)
assert not safe and len(generic) < len(reasons)
assert storage._canonical(original) == storage._canonical(current)
assert storage.revision_identity(original) == storage.revision_identity(current)
def test_projection_sanitizes_and_omits_safe_or_missing_reasons():
contract, safe, _ = signatures._sampling_contract(reroll_from_chunk=0)
assert safe
manifest = {"sampling_contract_version": 6, "resume_safe": True, "contract": contract,
"resume_disabled_reasons": list(adapter.PREVIEW_REUSE_REASONS)}
assert storage._project_resume_diagnostics(manifest) == {}
manifest["resume_safe"] = False
private = "C:/Users/private/file.py" + "x" * 2000
manifest["resume_disabled_reasons"] = [private] * 12
projected = storage._project_resume_diagnostics(manifest)["resume_disabled_reasons"]
assert projected == [storage._UNKNOWN_REUSE_REASON]
assert private not in json.dumps(projected) and max(map(len, projected)) <= 256
manifest.pop("resume_disabled_reasons")
assert storage._project_resume_diagnostics(manifest) == {}
def test_project_reason_belongs_to_its_saved_revision(tmp_path):
contract, safe, _ = signatures._sampling_contract(reroll_from_chunk=0)
assert safe
source = review._persist(tmp_path, contract, prefix=1, review_unit=(1, 1),
updated_utc="2026-10-19T00:00:00+00:00")
source.manifest["resume_safe"] = False
source.manifest["resume_disabled_reasons"] = ["KJ preview runtime code or state could not be verified"]
source._write_manifest()
manifest_before = source._manifest_path().read_bytes()
source._write_project()
project = review._project(source)
assert project["group_revisions"][0]["resume_disabled_reasons"] == source.manifest["resume_disabled_reasons"]
assert source._manifest_path().read_bytes() == manifest_before
@pytest.mark.parametrize("display_id", [None, "display-node"])
def test_public_node_uses_core_hidden_display_id(kj, monkeypatch, display_id):
has_display_mapping = "dynprompt" in kj.ModelPreviewOverrideKJ.execute.__code__.co_names
from comfy_api.latest import io
# Full-suite helpers can put the plugin ahead of Core; both contain nodes.py.
core_root = Path(kj.latent_preview.__file__).resolve().parent
monkeypatch.syspath_prepend(str(core_root))
from comfy_execution.graph import DynamicPrompt
import nodes as core_nodes
assert Path(core_nodes.__file__).resolve() == core_root / "nodes.py"
node_class = kj.ModelPreviewOverrideKJ
dynprompt = None
if display_id is not None:
dynprompt = DynamicPrompt({display_id: {}})
dynprompt.add_ephemeral_node("generated-node", {}, display_id, display_id)
hidden = io.HiddenHolder.from_dict({io.Hidden.unique_id: "generated-node", io.Hidden.dynprompt: dynprompt})
monkeypatch.setattr(node_class, "hidden", hidden)
class Model:
def clone(self):
return self
def add_wrapper_with_key(self, kind, key, value):
self.wrapper = value
model = Model()
node_class.execute(model, 512, 80, True, 1, 12)
assert model.wrapper.node_id == ((display_id or "generated-node") if has_display_mapping else "generated-node")
assert adapter.preview_override_signature(model.wrapper)[1]
+13
View File
@@ -0,0 +1,13 @@
import pytest
@pytest.mark.parametrize("case", [
"v6 read-only reports saved preview source and keeps actions disabled",
"preview setting remedies preserve generation VAE guidance",
"missing compatibility metadata is not mislabeled v5 or a known KJ error",
"unknown saved reasons do not expose paths or unbounded text",
"history reasons follow the selected Take rather than the canonical revision",
"saved compatibility metadata controls Review with optional real backend vectors",
])
def test_readonly_reason_with_real_frontend(review_queue_results, case):
assert review_queue_results[case]["pass"]
+1 -1
View File
@@ -6,7 +6,7 @@ import tomllib
ROOT = Path(__file__).resolve().parents[1]
EXPECTED_VERSION = "3.9.4"
EXPECTED_VERSION = "3.9.10"
def test_v35_release_metadata_is_consistent():
-124
View File
@@ -1,124 +0,0 @@
from __future__ import annotations
import json
import pytest
from ComfyUI_H3_Continuum_Join.prompt_builder import (
DumasH3SequencePromptBuilder,
default_sequence,
format_sequence_prompt,
normalize_sequence,
)
def test_builder_emits_one_absolute_timeline_prompt_from_clip_local_slots():
sequence = normalize_sequence(
{"clips": [
{"duration": 4, "slots": [{"start": 0, "end": 2, "text": "Walk forward."}]},
{"duration": 5, "slots": [{"start": 1, "end": 4, "text": "Turn toward the light."}]},
]},
clip_count=2,
default_seconds=4,
)
result = format_sequence_prompt(
shared_directive="Preserve the same character and environment.",
sequence=sequence,
soundscape="Quiet street ambience.",
music="Low cinematic pulse.",
)
assert "[0-4s]\n0-2s: First." in result
assert "[4-8s]\n1-4s: Turn toward the light." in result
assert "[0-2s]" not in result
assert "[5-9s]" not in result
assert "overall_soundscape: Quiet street ambience." in result
assert "non_diegetic_music: Low cinematic pulse." in result
assert result.count("Preserve the same character and environment.") == 2
assert result.count("overall_soundscape: Quiet street ambience.") == 2
assert result.count("non_diegetic_music: Low cinematic pulse.") == 2
def test_normalizer_orders_slots_fills_missing_clips_and_shares_duration():
value = normalize_sequence(
json.dumps({"clips": [{"duration": 6, "slots": [
{"start": 3, "end": 6, "text": "Second."},
{"start": 0, "end": 3, "text": "First."},
]}]}),
clip_count=2,
default_seconds=4,
)
assert value["clips"][0]["slots"][0]["text"] == "First."
assert value["clips"][0]["duration"] == 4.0
assert value["clips"][1]["duration"] == 4.0
@pytest.mark.parametrize("bad", [
{"clips": [{"duration": 4, "slots": [{"start": 3, "end": 2, "text": "bad"}]}]},
{"clips": [{"duration": 4, "slots": [{"start": 0, "end": 5, "text": "bad"}]}]},
])
def test_normalizer_rejects_invalid_slot_ranges(bad):
with pytest.raises(ValueError):
normalize_sequence(bad, clip_count=1, default_seconds=4)
def test_node_schema_has_prompt_outputs_reference_bundle_and_serialized_editor_state():
inputs = DumasH3SequencePromptBuilder.INPUT_TYPES()
schema = inputs["required"]
assert DumasH3SequencePromptBuilder.RETURN_TYPES == (
"STRING", "INT", "FLOAT", "H3_CONTINUUM_REFERENCE_IMAGES_V39"
)
assert "reference_image_1" in inputs["optional"]
assert schema["timeline_json"][1]["multiline"] is True
result = DumasH3SequencePromptBuilder().build(1, 4, "Directive", json.dumps(default_sequence(1, 4)), "N/A", "N/A")
assert result[0].startswith("[0-4s]\nDirective")
assert result[0].count("overall_soundscape:") == 1
assert result[1] == 1
assert result[2] == 4.0
assert result[3].reference_use == "Per chunk"
assert result[3].selectors == ("off",) * 9
def test_builder_routes_connected_references_only_to_mentioning_chunks():
value = {
"clips": [
{"slots": [{"start": 0, "end": 4, "text": "@R1 walks toward @R2."}]},
{"slots": [{"start": 0, "end": 4, "text": "The camera follows @image2."}]},
{"slots": [{"start": 0, "end": 4, "text": "A wide shot with no reference."}]},
]
}
result = DumasH3SequencePromptBuilder().build(
3,
4,
"",
json.dumps(value),
"",
"",
reference_image_1=object(),
reference_image_2=object(),
)
bundle = result[3]
assert bundle.reference_use == "Per chunk"
assert bundle.selectors == ("1", "1,2", "off", "off", "off", "off", "off", "off", "off")
assert bundle.images[0] is not None and bundle.images[1] is not None
def test_builder_routes_references_mentioned_anywhere_in_each_chunk_prompt():
value = {
"clips": [
{"slots": [{"start": 0, "end": 4, "text": "The subject walks."}]},
{"slots": [{"start": 0, "end": 4, "text": "The subject stops."}]},
]
}
result = DumasH3SequencePromptBuilder().build(
2,
4,
"Preserve the environment from @R1.",
json.dumps(value),
"Use ambience from @R2.",
"Music follows @R3.",
reference_image_1=object(),
reference_image_2=object(),
reference_image_3=object(),
)
assert result[3].selectors[:3] == ("1,2", "1,2", "1,2")
+104
View File
@@ -0,0 +1,104 @@
"""Preserve audited Spectrum identity across harmless newline/runtime differences."""
from __future__ import annotations
import hashlib
import importlib
import json
from pathlib import Path
import sys
import types
import pytest
from ComfyUI_H3_Continuum_Join import spectrum_signature as adapter
from .test_issue31_spectrum_signature import binding, signature, spectrum
def _relocate_provider(spectrum, tmp_path, monkeypatch, *, crlf=False):
folder = Path(spectrum[0].__file__).resolve().parent
for path in folder.glob("*.py"):
data = path.read_bytes()
if crlf:
data = data.replace(b"\r\n", b"\n").replace(b"\n", b"\r\n")
(tmp_path / path.name).write_bytes(data)
prefix = spectrum[0].__name__.rsplit(".", 1)[0]
for name, module in tuple(sys.modules.items()):
if name == prefix or name.startswith(prefix + "."):
original = getattr(module, "__file__", None)
if original and Path(original).resolve().parent == folder:
monkeypatch.setattr(module, "__file__", str(tmp_path / Path(original).name))
return prefix
def test_source_digest_normalizes_only_crlf(tmp_path):
path = tmp_path / "provider.py"
source = b"def value():\n return 'unchanged'\n"
path.write_bytes(source)
expected = hashlib.sha256(source).hexdigest()
assert adapter._source_digest(path) == expected
path.write_bytes(source.replace(b"\n", b"\r\n"))
assert adapter._source_digest(path) == expected
path.write_bytes(source.replace(b"unchanged", b"changed"))
assert adapter._source_digest(path) != expected
@pytest.mark.parametrize("newlines", ["all", "one"])
def test_crlf_preserves_full_existing_descriptor(spectrum, tmp_path, monkeypatch, newlines):
item = binding(spectrum)
before = signature(item)
_relocate_provider(spectrum, tmp_path, monkeypatch, crlf=newlines == "all")
if newlines == "one":
path = tmp_path / "config.py"
path.write_bytes(path.read_bytes().replace(b"\r\n", b"\n").replace(b"\n", b"\r\n"))
assert signature(item) == before
def test_module_closure_crlf_keeps_original_digest(spectrum, tmp_path, monkeypatch):
prefix = spectrum[0].__name__.rsplit(".", 1)[0]
module = importlib.import_module(prefix + ".runtime")
before = adapter._callable_value(module, provider_prefix=prefix)
_relocate_provider(spectrum, tmp_path, monkeypatch, crlf=True)
assert adapter._callable_value(module, provider_prefix=prefix) == before
@pytest.mark.parametrize("change", ["code", "extra_file"])
def test_real_source_changes_remain_unsafe(spectrum, tmp_path, monkeypatch, change):
item = binding(spectrum)
signature(item)
_relocate_provider(spectrum, tmp_path, monkeypatch, crlf=True)
if change == "code":
path = tmp_path / "config.py"
path.write_bytes(path.read_bytes() + b"\nUNKNOWN_ADDITIONAL_SETTING = True\n")
else:
(tmp_path / "unimported_helper.py").write_text("pass\n", encoding="utf-8")
descriptor, safe = adapter.spectrum_binding_signature(item)
assert not safe and descriptor.get("unsupported")
@pytest.mark.parametrize("version,suffix", [((3, 12, 10), "_py312"), ((3, 13, 12), "")])
def test_runtime_selects_explicit_audited_profile(monkeypatch, version, suffix):
monkeypatch.setattr(adapter, "sys", types.SimpleNamespace(version_info=version))
for name in ("spectrum_provider_v1.json", "spectrum_provider_v029.json"):
assert adapter._runtime_metadata_file(name) == Path(name).stem + suffix + ".json"
@pytest.mark.parametrize("version", [(3, 10, 0), (3, 11, 0), (3, 14, 0)])
def test_unaudited_python_runtime_stays_unsafe(monkeypatch, version):
monkeypatch.setattr(adapter, "sys", types.SimpleNamespace(version_info=version))
with pytest.raises(ValueError, match="unsupported provider Python runtime"):
adapter._runtime_metadata_file("spectrum_provider_v029.json")
def test_original_py313_profile_bytes_and_new_profile_packaging():
root = Path(adapter.__file__).parent
original = {
"spectrum_provider_v1.json": "0efc8d65cb6f59042457ad843fe1b77a4bee45e59d3732f5c83fbc75a31ae77b",
"spectrum_provider_v029.json": "8f786989af6339be8cec8f2bf602da89344cf96202d29c980af5c07175eea3d3",
}
for name, digest in original.items():
assert hashlib.sha256((root / name).read_bytes()).hexdigest() == digest
old = json.loads((root / name).read_text(encoding="utf-8"))
current = json.loads((root / (Path(name).stem + "_py312.json")).read_text(encoding="utf-8"))
assert old["format"] == current["format"] == "audited_spectrum_loaded_implementation_v1"
assert set(old["modules"]) == set(current["modules"])
assert old != current
-125
View File
@@ -1,125 +0,0 @@
import importlib.util
import pathlib
import unittest
MODULE_PATH = pathlib.Path(__file__).parents[1] / "v3" / "tiled_latent_upscale_nodes.py"
SPEC = importlib.util.spec_from_file_location("h3_tiled_test", MODULE_PATH)
MODULE = importlib.util.module_from_spec(SPEC)
assert SPEC.loader is not None
SPEC.loader.exec_module(MODULE)
class FakeSamples:
def __init__(self, shape):
self.shape = shape
self.ndim = len(shape)
class FakeNoise:
def get_noise(self, seed):
return ("noise", seed)
class FakeSampler:
def get_sampler(self, name):
return ("sampler", name)
class FakeScheduler:
def get_sigmas(self, model, scheduler, steps, denoise):
return ("sigmas", scheduler, steps, denoise)
class FakeNodeOutput:
def __init__(self, *args):
self.result = args
class FakeUpscale:
FUNCTION = "execute"
calls = []
def execute(self, **kwargs):
if "value" in kwargs:
return ({"samples": FakeSamples((1, 16, kwargs["value"], 1, 1))},)
self.calls.append(kwargs)
source = kwargs["latent"]["samples"]
_, _, frames, height, width = source.shape
return ({"samples": FakeSamples((1, 16, frames, height * 2, width * 2))},)
class TiledUpscaleTests(unittest.TestCase):
def setUp(self):
self.original = MODULE._native_node
FakeUpscale.calls = []
MODULE._native_node = lambda name: {
"MMH3UltimateUpscale": FakeUpscale,
"RandomNoise": FakeNoise,
"KSamplerSelect": FakeSampler,
"BasicScheduler": FakeScheduler,
}.get(name)
def tearDown(self):
MODULE._native_node = self.original
def test_each_clip_is_upscaled_independently(self):
node = MODULE.H3ContinuumTiledLatentUpscalePerClip()
clips = [
{"samples": FakeSamples((1, 16, 85, 48, 84))},
{"samples": FakeSamples((1, 16, 141, 48, 84))},
]
outputs, report = node.upscale(
model=["model"], clip=["clip"], video_latents=clips,
assembly_plan=["plan"], refine_context=["context"],
latent_upscale_param=["upscale"], spatial_split_param=["spatial"],
upscale_sampler=["euler"], upscale_scheduler=["simple"],
upscale_steps=[3], upscale_denoise=[0.2], upscale_cfg=[1.0],
upscale_noise_seed=[100], conditioning=["conditioning"],
)
self.assertEqual(len(outputs), 2)
self.assertEqual([call["latent"] for call in FakeUpscale.calls], clips)
self.assertEqual([call["noise"] for call in FakeUpscale.calls], [("noise", 100), ("noise", 101)])
self.assertIn("clips=2", report)
def test_native_function_name_is_taken_from_comfy_contract(self):
result = MODULE._call_native(FakeUpscale, {"value": 7})
self.assertEqual(result[0]["samples"].shape, (1, 16, 7, 1, 1))
def test_node_output_wrapper_is_unwrapped_for_single_values(self):
self.assertEqual(
MODULE._unwrap_single_output(FakeNodeOutput("tensor"), "schedule"),
"tensor",
)
self.assertEqual(
MODULE._unwrap_single_output(("legacy", "tuple"), "legacy"),
("legacy", "tuple"),
)
def test_rejects_temporal_resize(self):
class BadUpscale(FakeUpscale):
def execute(self, **kwargs):
source = kwargs["latent"]["samples"]
return ({"samples": FakeSamples((1, 16, source.shape[2] + 1, 96, 168))},)
MODULE._native_node = lambda name: {
"MMH3UltimateUpscale": BadUpscale,
"RandomNoise": FakeNoise,
"KSamplerSelect": FakeSampler,
"BasicScheduler": FakeScheduler,
}.get(name)
node = MODULE.H3ContinuumTiledLatentUpscalePerClip()
with self.assertRaisesRegex(RuntimeError, "temporal length"):
node.upscale(
model=["model"], clip=["clip"],
video_latents=[{"samples": FakeSamples((1, 16, 85, 48, 84))}],
assembly_plan=["plan"], refine_context=["context"],
latent_upscale_param=["upscale"], spatial_split_param=["spatial"],
upscale_sampler=["euler"], upscale_scheduler=["simple"],
upscale_steps=[3], upscale_denoise=[0.2], upscale_cfg=[1.0],
upscale_noise_seed=[100], conditioning=["conditioning"],
)
if __name__ == "__main__":
unittest.main()
+1 -1
View File
@@ -145,4 +145,4 @@ def test_v217_result_pack_expands_legacy_outputs():
def test_v217_has_no_frontend_display_controller():
assert not (ROOT / "web" / "h3_continuum_v2.js").exists()
version_source = (ROOT / "version.py").read_text(encoding="utf-8")
assert 'PACKAGE_VERSION = "3.9.4"' in version_source
assert 'PACKAGE_VERSION = "3.9.10"' in version_source
+2
View File
@@ -89,6 +89,8 @@ def test_run_storage_off_normalizes_widgets_reload_and_queued_inputs(tmp_path):
source = PROJECT_ID_JS.read_text(encoding="utf-8")
functions = "\n".join(
[
next(line for line in source.splitlines()
if line.startswith("const REGENERATE_INTENT_PROPERTY = ")),
_function_source(source, "findWidget", "setWidgetVisible"),
_function_source(source, "setWidgetVisible", "hidePersistentWidget"),
_function_source(source, "linkedInput", "attachRefresh"),
-93
View File
@@ -543,99 +543,6 @@ def test_second_pass_node_forwards_optional_refine_context_and_video_vae(monkeyp
assert captured["refine_schedule"] is schedule
def test_tiled_second_pass_prefers_connected_sigma_socket_over_node_steps(monkeypatch):
captured = {}
class FakeNative:
pass
class FakeNoise:
pass
class FakeSampler:
def get_sampler(self, name):
return f"sampler:{name}"
class FakeScheduler:
pass
def fake_run(**kwargs):
captured.update(kwargs)
return ([], [], {"plan": True}, "status")
monkeypatch.setattr(
"ComfyUI_H3_Continuum_Join.v3.second_pass_nodes.run_second_pass_groups",
fake_run,
)
monkeypatch.setattr(
"ComfyUI_H3_Continuum_Join.v3.second_pass_nodes._native_node",
lambda name: {
"MMH3UltimateUpscale": FakeNative,
"RandomNoise": FakeNoise,
"KSamplerSelect": FakeSampler,
"BasicScheduler": FakeScheduler,
}[name],
)
socket_sigmas = object()
result = H3ContinuumSecondPassV35().refine(
["model"], ["clip"], ["sampler"], [socket_sigmas],
[], [], [{"second_pass_contract": {}}], [7],
tiled_upscale=[True], latent_upscale_param=["upscale"],
spatial_split_param=["spatial"], upscale_steps=[3],
upscale_denoise=[0.2], upscale_cfg=[1.0], upscale_noise_seed=[11],
)
assert result[0] == []
config = captured["tiled_upscale_config"]
assert config["sigmas"] is socket_sigmas
assert config["sigma_source"] == "sigmas_socket"
def test_tiled_second_pass_prefers_dedicated_sigma_socket(monkeypatch):
captured = {}
class FakeNative:
pass
class FakeNoise:
pass
class FakeSampler:
def get_sampler(self, name):
return name
class FakeScheduler:
pass
monkeypatch.setattr(
"ComfyUI_H3_Continuum_Join.v3.second_pass_nodes.run_second_pass_groups",
lambda **kwargs: captured.update(kwargs) or ([], [], {}, ""),
)
monkeypatch.setattr(
"ComfyUI_H3_Continuum_Join.v3.second_pass_nodes._native_node",
lambda name: {
"MMH3UltimateUpscale": FakeNative,
"RandomNoise": FakeNoise,
"KSamplerSelect": FakeSampler,
"BasicScheduler": FakeScheduler,
}[name],
)
main_sigmas = object()
tiled_sigmas = object()
H3ContinuumSecondPassV35().refine(
["model"], ["clip"], ["sampler"], [main_sigmas],
[], [], [{"second_pass_contract": {}}], [7],
tiled_upscale=[True], latent_upscale_param=["upscale"],
spatial_split_param=["spatial"], upscale_sigmas=[tiled_sigmas],
upscale_steps=[3], upscale_denoise=[0.2], upscale_cfg=[1.0],
upscale_noise_seed=[11],
)
config = captured["tiled_upscale_config"]
assert config["sigmas"] is tiled_sigmas
assert config["sigma_source"] == "upscale_sigmas_socket"
def test_second_pass_node_is_registered_and_not_deprecated():
node_class = NODE_CLASS_MAPPINGS["H3ContinuumSecondPassV35"]
assert node_class is H3ContinuumSecondPassV35
+3
View File
@@ -194,6 +194,9 @@ def test_registry_manifest_matches_every_declared_source_file():
"examples/workflows/MiniMax_H3_Continuum_V39.json",
"examples/workflows/MiniMax_H3_Continuum_V39.zip",
"examples/MiniMax_H3_Continuum_V39.json",
"examples/MiniMax_H3_Continuum_V39_Turbo_Sample.json",
"examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.json",
"examples/workflows/MiniMax_H3_Continuum_V39_Turbo_Sample.zip",
"examples/MiniMax_H3_Continuum_V38X2.json",
}
+102 -2
View File
@@ -13,6 +13,7 @@ import pytest
from ComfyUI_H3_Continuum_Join.v3.driving_nodes import (
H3ContinuumSamplerV37,
H3ContinuumSamplerV38,
H3ContinuumSamplerV39,
)
from ComfyUI_H3_Continuum_Join.v3.easy_nodes import H3ContinuumEasyV38
from ComfyUI_H3_Continuum_Join.v3.review_control import (
@@ -28,7 +29,7 @@ ROOT = Path(__file__).resolve().parents[1]
DEFAULT_COMFY_ROOT = Path(
os.environ.get(
"H3_CONTINUUM_COMFY_ROOT",
r"D:\StabilityMatrix\Data\Packages\ComfyUI_W",
str(ROOT.parent / "ComfyUI"),
)
)
@@ -71,6 +72,25 @@ def test_v38_full_run_keeps_stable_core_cache_behavior(review_action):
) is False
@pytest.mark.parametrize("node_class", [H3ContinuumSamplerV38, H3ContinuumSamplerV39])
@pytest.mark.parametrize("boundary,nonce,storage,changed", [
("Chunk 1", 0, "Save + Auto Resume", True),
("Chunk 2", 0, "Save + Auto Resume", True),
(2, 0, "Save + Auto Resume", True),
("Chunk 2", 5, "Save + Auto Resume", False),
("Auto", 0, "Save + Auto Resume", False),
(0, 0, "Save + Auto Resume", False),
("Chunk 2", 0, "Off", False),
])
def test_full_run_automatic_reroll_reaches_storage(node_class, boundary, nonce, storage, changed):
value = node_class.IS_CHANGED(generation_mode=GENERATION_MODE_FULL_RUN,
run_storage=storage, reroll_from_chunk=boundary, reroll_nonce=nonce, chunks=3)
if changed:
assert math.isnan(value)
else:
assert value is False
def test_v37_and_easy_do_not_inherit_the_v38_cache_contract():
assert "IS_CHANGED" not in H3ContinuumSamplerV37.__dict__
assert "IS_CHANGED" not in H3ContinuumEasyV38.__dict__
@@ -78,6 +98,37 @@ def test_v37_and_easy_do_not_inherit_the_v38_cache_contract():
assert not hasattr(H3ContinuumEasyV38, "IS_CHANGED")
@pytest.mark.parametrize("node_class", [H3ContinuumSamplerV38, H3ContinuumSamplerV39])
@pytest.mark.parametrize("changes,changed", [
({"chunks": None}, True),
({"chunks": 1}, True),
({"chunks": "not resolved"}, True),
({"reroll_nonce": None}, True),
({"run_storage": None}, True),
({"reroll_from_chunk": None}, True),
({"generation_mode": None}, True),
({"chunks": None, "reroll_nonce": None, "run_storage": None}, True),
({"chunks": None, "reroll_nonce": 5}, False),
({"run_storage": None, "reroll_nonce": 5}, False),
({"reroll_from_chunk": None, "reroll_nonce": 5}, False),
({"reroll_from_chunk": "Auto", "reroll_nonce": None}, False),
({"reroll_from_chunk": "Auto", "run_storage": None}, False),
({"run_storage": "Off", "reroll_nonce": None}, False),
({"run_storage": "Off", "reroll_from_chunk": None}, False),
({"run_storage": "invalid"}, False),
({"reroll_nonce": "invalid"}, False),
({"reroll_from_chunk": "invalid"}, False),
({"reroll_from_chunk": "Chunk 17"}, False),
({"generation_mode": GENERATION_MODE_REVIEW, "run_storage": "Off"}, True),
])
def test_cache_intent_handles_unresolved_inputs_without_validating_chunk_range(node_class, changes, changed):
inputs = dict(generation_mode=GENERATION_MODE_FULL_RUN, run_storage="Save + Auto Resume",
reroll_from_chunk="Chunk 2", reroll_nonce=0, chunks=3)
inputs.update(changes)
value = node_class.IS_CHANGED(**inputs)
assert math.isnan(value) if changed else value is False
def test_real_comfy_cache_signature_is_stable_only_for_full_run(tmp_path):
comfy_root = DEFAULT_COMFY_ROOT
if not (comfy_root / "execution.py").is_file():
@@ -130,7 +181,7 @@ def test_real_comfy_cache_signature_is_stable_only_for_full_run(tmp_path):
comfy_kitchen.int8_attention_is_available = lambda *a, **k: False
import nodes as comfy_nodes
from comfy_execution.caching import CacheKeySetInputSignature
from comfy_execution.caching import BasicCache, CacheKeySetInputSignature
from comfy_execution.graph import DynamicPrompt
from execution import IsChangedCache
@@ -187,7 +238,49 @@ def test_real_comfy_cache_signature_is_stable_only_for_full_run(tmp_path):
review_finish_2 = await signature(
"Review Each Chunk", "Finish Remaining"
)
cache_results = {{}}
for node_name in ["H3ContinuumSamplerV38", "H3ContinuumSamplerV39"]:
comfy_nodes.NODE_CLASS_MAPPINGS[class_type] = package.NODE_CLASS_MAPPINGS[node_name]
observed_counts = {{}}
for label, boundary, nonce, linked in [
("ordinary", "Auto", 0, None),
("automatic", "Chunk 2", 0, None),
("fixed", "Chunk 2", 5, None),
("linked_chunks", "Chunk 2", 0, "chunks"),
("linked_nonce", "Chunk 2", 0, "reroll_nonce"),
("linked_storage", "Chunk 2", 0, "run_storage"),
("linked_boundary", "Chunk 2", 0, "reroll_from_chunk"),
("linked_mode", "Chunk 2", 5, "generation_mode"),
("fixed_linked_chunks", "Chunk 2", 5, "chunks"),
("ordinary_linked_nonce", "Auto", 0, "reroll_nonce"),
("fixed_linked_nonce", "Chunk 2", 5, "reroll_nonce"),
]:
cache = BasicCache(CacheKeySetInputSignature)
executions = 0
for queue in range(2):
prompt = {{"305": {{"class_type": class_type, "inputs": {{
"generation_mode": "Full Run", "review_action": "Continue / Next",
"run_storage": "Save + Auto Resume", "reroll_from_chunk": boundary,
"reroll_nonce": nonce, "chunks": 3}}}}}}
if linked:
class ScalarProbe:
@classmethod
def INPUT_TYPES(cls):
return {{"required": {{"value": ("*",)}}}}
comfy_nodes.NODE_CLASS_MAPPINGS["H3CacheScalarProbe"] = ScalarProbe
value = prompt["305"]["inputs"][linked]
prompt["900"] = {{"class_type": "H3CacheScalarProbe", "inputs": {{"value": value}}}}
prompt["305"]["inputs"][linked] = ["900", 0]
dynprompt = DynamicPrompt(prompt)
await cache.set_prompt(dynprompt, ["305"], IsChangedCache(str(queue), dynprompt, None))
if cache.get_local("305") is None:
executions += 1
cache.set_local("305", ("CPU fixture output",))
observed_counts[label] = executions
cache_results[node_name] = observed_counts
print(json.dumps({{
"cache_executions": cache_results,
"full_stable": full_1 == full_2,
"review_continue_changed": review_continue_1 != review_continue_2,
"review_regenerate_changed": review_regen_1 != review_regen_2,
@@ -210,6 +303,13 @@ def test_real_comfy_cache_signature_is_stable_only_for_full_run(tmp_path):
)
observed = json.loads(result.stdout.strip().splitlines()[-1])
assert observed == {
"cache_executions": {
name: {"ordinary": 1, "automatic": 2, "fixed": 1,
"linked_chunks": 2, "linked_nonce": 2, "linked_storage": 2,
"linked_boundary": 2, "linked_mode": 2, "fixed_linked_chunks": 1,
"ordinary_linked_nonce": 1, "fixed_linked_nonce": 2}
for name in ("H3ContinuumSamplerV38", "H3ContinuumSamplerV39")
},
"full_stable": True,
"review_continue_changed": True,
"review_regenerate_changed": True,
+8
View File
@@ -471,3 +471,11 @@ def test_n2c_history_refresh_hooks_and_multiline_surface_are_present():
assert 'const TAKE_TOGGLE_WIDGET = "Render History";' in source
assert "app.queuePrompt =" not in source
assert "api.__h3ContinuumReviewQueueAdapter" in source
def test_explicit_regenerate_restore_and_legacy_compatibility(review_queue_results):
selected = [record for name, record in review_queue_results.items()
if name.startswith(("regenerate restore:", "regenerate legacy:",
"regenerate intent reset:"))]
assert len(selected) == 22
assert all(record["pass"] for record in selected)
+1
View File
@@ -98,6 +98,7 @@ def test_template_gallery_uses_official_workflows_only():
gallery = ROOT / "examples"
assert {path.name for path in gallery.glob("*.json")} == {
"MiniMax_H3_Continuum_V39.json",
"MiniMax_H3_Continuum_V39_Turbo_Sample.json",
"MiniMax_H3_Continuum_V38X2.json",
}
assert (gallery / WORKFLOW.name).read_bytes() == WORKFLOW.read_bytes()
+22
View File
@@ -177,6 +177,28 @@ def test_inline_helper_keeps_sparse_slot_ids_and_saved_future_chunks():
assert bundle.selectors[0] == "1,3,6"
@pytest.mark.parametrize("saved,expected", [
("1,2", [("R1",), ("R1",), ()]),
("1,2,3", [("R1",), ("R1",), ("R1",)]),
(",".join(str(n) for n in range(1, 17)), [("R1",), ("R1",), ("R1",)]),
("off", [(), (), ()]),
])
def test_external_assignment_range_uses_actual_chunks_without_rewriting_saved_value(saved, expected):
bundle, = H3ContinuumReferenceImagesV39().pack(
reference_use="Per chunk", reference_r1_chunks=saved,
reference_image_1=object(),
)
original = bundle.selectors
_, _, selectors = effective_reference_inputs(bundle, chunks=3)
schedule = compile_reference_routing_schedule(
total_chunks=3, terminal_merge_enabled=False,
mode="Custom", selectors_by_slot=selectors,
)
assert schedule.valid
assert [route.reference_slot_ids for route in schedule.logical_routes] == expected
assert bundle.selectors == original
def test_new_workflow_rewires_all_nine_slots_and_isolates_saved_runs():
directory = Path(__file__).resolve().parents[1] / "examples" / "workflows"
old = json.loads((directory / "MiniMax_H3_Continuum_V38X2.json").read_text(encoding="utf-8"))
+36
View File
@@ -0,0 +1,36 @@
import json
from pathlib import Path
from zipfile import ZipFile
ROOT = Path(__file__).resolve().parents[1]
NAME = "MiniMax_H3_Continuum_V39_Turbo_Sample.json"
def test_turbo_sample_is_installable_and_keeps_prompt_and_explicit_settings():
path = ROOT / "examples/workflows" / NAME
payload = path.read_bytes()
assert (ROOT / "examples" / NAME).read_bytes() == payload
with ZipFile(path.with_suffix(".zip")) as archive:
assert archive.testzip() is None
assert archive.namelist() == [NAME]
assert archive.read(NAME) == payload
graph = json.loads(payload)
nodes = {n["id"]: n for n in graph["nodes"]}
named = nodes[312]["widgets_values_named"]
assert nodes[312]["widgets_values"] == list(named.values())
assert (named["prompt_mode"], named["chunks"], named["chunk_seconds"]) == ("Auto", 2, 8)
assert named["run_storage"] == "Off"
assert named["project_id"] != "1fefd563-68c4-45f7-a33f-8a2911850a9c"
assert nodes[317]["mode"] == 0
assert next(i for i in nodes[312]["inputs"] if i["name"] == "first_frame")["link"] == 382
prompt = nodes[188]["widgets_values"][0]
assert prompt == nodes[188]["widgets_values_named"]["value"]
assert prompt.startswith("[Chunk 1]\nFor the target video")
assert "<Picture 1>" in prompt and "[Chunk 2]\n" in prompt
assert "Continuation of Chunk 1." in prompt
assert prompt.count("<d>[Thai]") == 2
assert "background unchanged" in prompt
enabled = [v for v in nodes[286]["widgets_values"] if isinstance(v, dict) and v.get("on")]
assert len(enabled) == 1 and "turbo_4step" in enabled[0]["lora"]
assert nodes[150]["widgets_values"][0] is False
assert "models" not in nodes[147]["properties"]
assert nodes[191]["widgets_values"][0].endswith("_Turbo_Sample")
+27 -7
View File
@@ -61,7 +61,11 @@ from .memory_action_policy import (
resolve_memory_action_policy,
)
from .node_facade import execute_v38_runtime_request, make_runtime_request
from .nodes import CATEGORY as CONTINUUM_CATEGORY, H3ContinuumSamplerProduction
from .nodes import (
CATEGORY as CONTINUUM_CATEGORY,
H3ContinuumSamplerProduction,
_regenerate_from_value,
)
from .resolution import (
H3_ASPECT_AUTO,
H3_ASPECT_OPTIONS,
@@ -739,15 +743,31 @@ class H3ContinuumSamplerV38(H3ContinuumSamplerV37):
def IS_CHANGED(
cls,
generation_mode=GENERATION_MODE_FULL_RUN,
run_storage="Off",
reroll_from_chunk="Auto",
reroll_nonce=0,
**kwargs,
):
# Review execution depends on validated Run Storage state that is
# intentionally outside the visible graph inputs. Always execute the
# V3.8 sampler in Review mode and let Run Storage remain the sole
# authority for exact prefix reuse. Full Run keeps Core's stable cache
# behavior, and this method is deliberately not inherited by V3.7/Easy.
if generation_mode == GENERATION_MODE_REVIEW:
# Review and automatic reroll variations depend on saved state outside
# Core's input cache. Run Storage still owns all reuse/safety decisions.
# Core supplies None for linked inputs during its cache-key pass.
if generation_mode is None or generation_mode == GENERATION_MODE_REVIEW:
return float("NaN")
if run_storage in ("Save + Auto Resume", None):
try:
if reroll_nonce is not None and int(reroll_nonce) != 0:
return False
if reroll_from_chunk is None:
return float("NaN")
# The execution path validates the actual configured chunk range.
boundary = _regenerate_from_value(
reroll_from_chunk, chunks=16
)
if boundary > 0:
return float("NaN")
except (TypeError, ValueError, OverflowError):
pass # Core/backend validation owns malformed inputs.
# Ordinary Full Run and fixed positive nonces keep stable caching.
return False
def run(
+20 -103
View File
@@ -29,7 +29,6 @@ from .refine_schedule import (
serializable_schedule_contract,
)
from .refine_sampling import sample_refine_chunk
from .tiled_latent_upscale_nodes import _call_native, _unwrap_single_output
class SecondPassContractError(ValueError):
@@ -458,7 +457,6 @@ def run_second_pass_groups(
refine_schedule: RefineSchedule | None = None,
selected_group_indices: Sequence[int] | None = None,
temporal_windows_by_group: Mapping[int, Any] | None = None,
tiled_upscale_config: Mapping[str, Any] | None = None,
) -> tuple[list[dict[str, Any]], list[Any], dict[str, Any], str]:
"""Refine complete physical groups while preserving first-pass audio exactly."""
@@ -503,7 +501,6 @@ def run_second_pass_groups(
context_frames = int(detail["context_frames"])
physical_seed = derive_refine_seed(int(refine_seed), group_index)
group_seeds.append(physical_seed)
upscale_seed = None
conditioning_sources.append(conditioning_source)
route_record = detail.get("reference_routing_v1")
inheritance = (
@@ -517,74 +514,27 @@ def run_second_pass_groups(
"contract_sha256": route_record.get("sha256") if route_record is not None else None,
})
nested_latent = latent_builder(video_samples, audio_samples)
if tiled_upscale_config is None:
sample_arguments = dict(
model=chunk_model,
conditioning=conditioning,
latent=nested_latent,
sampler=sampler,
sigmas=effective_sigmas,
seed=physical_seed,
enable_preview=bool(enable_preview),
)
if temporal_windows_by_group is not None:
temporal_window = temporal_windows_by_group.get(group_index)
if temporal_window is None:
raise SecondPassContractError(
f"selected physical group {group_index + 1} has no temporal window"
)
sample_arguments["temporal_window"] = temporal_window
sampled = sample_fn(**sample_arguments)
else:
config = tiled_upscale_config
native = config["native"]
upscale_seed = derive_refine_seed(
int(config.get("seed_base", physical_seed)), group_index
)
noise = _unwrap_single_output(
config["random_noise"]().get_noise(upscale_seed), "RandomNoise"
)
upscale_sampler = config["sampler"]
upscale_sigmas = config.get("sigmas")
if upscale_sigmas is None:
upscale_sigmas = _unwrap_single_output(
config["scheduler"]().get_sigmas(
chunk_model,
config["scheduler_name"],
config["steps"],
config["denoise"],
),
"BasicScheduler",
)
native_kwargs = {
"model": chunk_model,
"conditioning": conditioning,
"latent": nested_latent,
"noise": noise,
"sampler": upscale_sampler,
"sigmas": upscale_sigmas,
"cfg": config["cfg"],
"latent_upscale_param": config["latent_upscale_param"],
"temporal_split_param": config.get("temporal_split_param"),
"spatial_split_param": config["spatial_split_param"],
}
if config.get("negative") is not None:
native_kwargs["negative"] = config["negative"]
native_result = _call_native(native, native_kwargs)
if not isinstance(native_result, (tuple, list)) or not native_result:
sample_arguments = dict(
model=chunk_model,
conditioning=conditioning,
latent=nested_latent,
sampler=sampler,
sigmas=effective_sigmas,
seed=physical_seed,
enable_preview=bool(enable_preview),
)
if temporal_windows_by_group is not None:
temporal_window = temporal_windows_by_group.get(group_index)
if temporal_window is None:
raise SecondPassContractError(
f"tiled upscale returned no output for physical group {group_index + 1}"
f"selected physical group {group_index + 1} has no temporal window"
)
sampled = native_result[0]
sample_arguments["temporal_window"] = temporal_window
sampled = sample_fn(**sample_arguments)
refined_video, _temporary_audio = stream_extractor(sampled)
if tiled_upscale_config is None:
if tuple(refined_video.shape) != tuple(video_samples.shape):
raise SecondPassContractError(
f"refined video group {group_index} changed B/C/T/H/W geometry"
)
elif tuple(refined_video.shape[:3]) != tuple(video_samples.shape[:3]):
if tuple(refined_video.shape) != tuple(video_samples.shape):
raise SecondPassContractError(
f"tiled upscale group {group_index} changed B/C/T geometry"
f"refined video group {group_index} changed B/C/T/H/W geometry"
)
if not bool(torch.isfinite(refined_video.float()).all().item()):
raise SecondPassContractError(
@@ -604,17 +554,9 @@ def run_second_pass_groups(
f"reference_inheritance={inheritance}, "
f"physical_clip_index={physical_clip_index}, "
f"context_frames={context_frames}, "
f"refine_seed={physical_seed}, "
f"upscale_seed={upscale_seed if upscale_seed is not None else 'n/a'}, "
f"shape={tuple(refined_video.shape)}, "
f"sampling_passes=1, tiled_upscale={tiled_upscale_config is not None}, "
"temporary_audio_discarded=true"
f"refine_seed={physical_seed}, shape={tuple(refined_video.shape)}, "
"sampling_passes=1, temporary_audio_discarded=true"
)
if tiled_upscale_config is not None:
group_report_lines.append(
"tiled sigma source="
f"{tiled_upscale_config.get('sigma_source', 'node_steps_fallback')}"
)
if adaptation_stats is not None:
group_report_lines.append(
"group "
@@ -677,36 +619,11 @@ def run_second_pass_groups(
passthrough_audio = passthrough_audio_latents(audio_latents)
updated_plan = prepared["updated_assembly_plan"]
if tiled_upscale_config is not None:
if selection_active:
raise SecondPassContractError(
"tiled upscale mode requires all physical groups; selective tiled "
"upscale would produce mixed output geometry"
)
output_samples = _latent_samples(
refined_videos[0], name="tiled refined video group 0"
)
target_height = int(output_samples.shape[-2]) * 16
target_width = int(output_samples.shape[-1]) * 16
updated_plan = copy.deepcopy(updated_plan)
updated_plan["height"] = target_height
updated_plan["width"] = target_width
updated_plan["second_pass_contract"]["target_height"] = target_height
updated_plan["second_pass_contract"]["target_width"] = target_width
for group in updated_plan["second_pass_contract"]["physical_groups"]:
group["target_latent_h"] = int(output_samples.shape[-2])
group["target_latent_w"] = int(output_samples.shape[-1])
group["target_height"] = target_height
group["target_width"] = target_width
contract = updated_plan["second_pass_contract"]
if conditioning_sources and all(
source == "refine_context" for source in conditioning_sources
):
contract["execution"] = (
"context_aware_physical_groups_tiled_upscale_audio_locked_v1"
if tiled_upscale_config is not None
else "context_aware_physical_groups_audio_locked_v3"
)
contract["execution"] = "context_aware_physical_groups_audio_locked_v3"
elif any(source == "refine_context" for source in conditioning_sources):
contract["execution"] = "mixed_context_physical_groups_audio_locked_v3"
else:
+1 -119
View File
@@ -9,7 +9,6 @@ from .refine_scope import MODE_ALL, MODE_LOGICAL_CHUNK, MODE_PHYSICAL_GROUP
from .refine_window import MODE_TIME_WINDOW
from .second_pass import run_second_pass_groups
from .targeted_second_pass import run_targeted_second_pass_groups
from .tiled_latent_upscale_nodes import _native_node, _unwrap_single_output
REFINE_TARGET_VIDEO_ONLY = "Video Only"
@@ -50,30 +49,6 @@ def _single_list_input(name: str, value: Any) -> Any:
return value[0]
def _optional_single_list_input(name: str, value: Any) -> Any:
"""Unwrap an optional INPUT_IS_LIST value without losing an absent socket."""
if value is None:
return None
return _single_list_input(name, value)
def _sampler_options():
try:
import comfy.samplers
return list(comfy.samplers.KSampler.SAMPLERS)
except Exception:
return ["euler", "euler_ancestral", "res_multistep", "sa_solver"]
def _scheduler_options():
try:
import comfy.samplers
return list(comfy.samplers.KSampler.SCHEDULERS)
except Exception:
return ["simple", "beta", "sgm_uniform", "karras", "exponential"]
class H3ContinuumSecondPassV35:
"""Low-sigma refinement over complete physical decode groups."""
@@ -84,10 +59,7 @@ class H3ContinuumSecondPassV35:
"Connect the optional Continuum refine context to preserve the physical group's "
"First/Last/Reference conditioning. "
"SIGMAS controls the low-denoise refine schedule; audio output is preserved "
"bit-exact from the first pass. In tiled-upscale mode, connected "
"upscale-specific sockets take priority; the main SIGMAS socket is the "
"fallback external schedule. The other optional tiled sockets fall back "
"to their node widgets when they are not connected."
"bit-exact from the first pass."
)
@classmethod
@@ -111,28 +83,10 @@ class H3ContinuumSecondPassV35:
"tooltip": "Base seed independently derived once per physical group.",
},
),
"tiled_upscale": (
"BOOLEAN",
{
"default": False,
"tooltip": "Use native tiled H3 latent upscale instead of the low-sigma refine pass.",
},
),
},
"optional": {
"refine_context": ("H3_CONTINUUM_REFINE_CONTEXT",),
"video_vae": ("VAE",),
"latent_upscale_param": ("H3_UPSCALE_PARAM",),
"spatial_split_param": ("H3_SPATIAL_PARAM",),
"temporal_split_param": ("H3_TEMPORAL_PARAM",),
"upscale_sampler": (_sampler_options(), {"default": "euler_ancestral"}),
"upscale_scheduler": (_scheduler_options(), {"default": "simple"}),
"upscale_steps": ("INT", {"default": 3, "min": 1, "max": 50}),
"upscale_denoise": ("FLOAT", {"default": 0.2, "min": 0.0, "max": 1.0, "step": 0.01}),
"upscale_cfg": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 20.0, "step": 0.1}),
"upscale_noise_seed": ("INT", {"default": 0, "min": 0, "max": 0xFFFFFFFFFFFFFFFF, "control_after_generate": True}),
"upscale_negative": ("CONDITIONING",),
"upscale_sigmas": ("SIGMAS",),
},
}
@@ -161,78 +115,7 @@ class H3ContinuumSecondPassV35:
refine_context=None,
video_vae=None,
refine_schedule=None,
tiled_upscale=False,
latent_upscale_param=None,
spatial_split_param=None,
temporal_split_param=None,
upscale_sampler="euler_ancestral",
upscale_scheduler="simple",
upscale_steps=3,
upscale_denoise=0.2,
upscale_cfg=1.0,
upscale_noise_seed=0,
upscale_negative=None,
upscale_sigmas=None,
):
tiled_config = None
if bool(_single_list_input("tiled_upscale", tiled_upscale)):
required_native = {
name: _native_node(name)
for name in (
"MMH3UltimateUpscale",
"RandomNoise",
"KSamplerSelect",
"BasicScheduler",
)
}
missing = [name for name, value in required_native.items() if value is None]
if missing:
raise RuntimeError("Missing native tiled-upscale nodes: " + ", ".join(missing))
if latent_upscale_param is None or spatial_split_param is None:
raise ValueError(
"Tiled upscale mode requires latent_upscale_param and spatial_split_param"
)
tiled_config = {
"native": required_native["MMH3UltimateUpscale"],
"random_noise": required_native["RandomNoise"],
"sampler": _unwrap_single_output(
required_native["KSamplerSelect"]().get_sampler(
str(_single_list_input("upscale_sampler", upscale_sampler))
),
"KSamplerSelect",
),
"scheduler": required_native["BasicScheduler"],
"scheduler_name": str(_single_list_input("upscale_scheduler", upscale_scheduler)),
"steps": int(_single_list_input("upscale_steps", upscale_steps)),
"denoise": float(_single_list_input("upscale_denoise", upscale_denoise)),
"cfg": float(_single_list_input("upscale_cfg", upscale_cfg)),
"seed_base": int(_single_list_input("upscale_noise_seed", upscale_noise_seed)),
"latent_upscale_param": _single_list_input("latent_upscale_param", latent_upscale_param),
"spatial_split_param": _single_list_input("spatial_split_param", spatial_split_param),
"temporal_split_param": (
None if temporal_split_param is None
else _single_list_input("temporal_split_param", temporal_split_param)
),
"negative": (
None if upscale_negative is None
else _single_list_input("upscale_negative", upscale_negative)
),
# A dedicated tiled schedule wins. The main SIGMAS socket is
# the backwards-compatible external fallback used by existing
# DUMAS workflows.
"sigmas": _optional_single_list_input(
"upscale_sigmas", upscale_sigmas
)
if upscale_sigmas is not None
else _optional_single_list_input("sigmas", sigmas),
"sigma_source": (
"upscale_sigmas_socket"
if upscale_sigmas is not None
else "sigmas_socket"
if sigmas is not None
else "node_steps_fallback"
),
}
return run_second_pass_groups(
model=_single_list_input("model", model),
clip=_single_list_input("clip", clip),
@@ -253,7 +136,6 @@ class H3ContinuumSecondPassV35:
else _single_list_input("video_vae", video_vae)
),
refine_schedule=refine_schedule,
tiled_upscale_config=tiled_config,
)
-357
View File
@@ -1,357 +0,0 @@
"""Per-clip tiled latent upscaling for MiniMax H3 Continuum.
Continuum exposes physical video groups as a LATENT list. The native
``MMH3UltimateUpscale`` node operates on one H3 latent at a time, so this
adapter deliberately applies it independently to each physical clip instead
of concatenating the clips into one larger VRAM allocation.
"""
from __future__ import annotations
from typing import Any
def _sampler_options() -> list[str]:
try:
import comfy.samplers
return list(comfy.samplers.KSampler.SAMPLERS)
except Exception: # pragma: no cover - exercised on a ComfyUI host
return ["euler", "euler_ancestral", "res_multistep", "sa_solver"]
def _scheduler_options() -> list[str]:
try:
import comfy.samplers
return list(comfy.samplers.KSampler.SCHEDULERS)
except Exception: # pragma: no cover - exercised on a ComfyUI host
return ["simple", "beta", "sgm_uniform", "karras", "exponential"]
def _native_node(name: str):
try:
import nodes
return (getattr(nodes, "NODE_CLASS_MAPPINGS", {}) or {}).get(name)
except Exception:
return None
def _unwrap_node_output(value: Any) -> Any:
"""Extract values returned by ComfyUI's newer NodeOutput API.
The v3 API wraps node results in ``NodeOutput`` while legacy nodes expect
the underlying tensor/object directly. Keep this adapter compatible with
both APIs without importing ComfyUI internals.
"""
result = getattr(value, "result", None)
if result is None and not hasattr(value, "result"):
return value
return result
def _unwrap_single_output(value: Any, name: str) -> Any:
marker = object()
result = getattr(value, "result", marker)
# Ordinary tuples are valid values for the lightweight test doubles and
# for some legacy ComfyUI helpers. Only unwrap an actual NodeOutput-like
# object, identified by its result property.
if result is marker:
return value
if isinstance(result, (tuple, list)):
if len(result) != 1:
raise ValueError(f"{name} returned {len(result)} outputs; expected one")
return result[0]
return result
def _call_native(node_class, kwargs: dict[str, Any]):
"""Invoke a ComfyUI node through its declared FUNCTION contract.
ComfyUI node classes do not have a universal Python method name. The live
MiniMax node currently declares its executor through ``FUNCTION`` rather
than exposing a method named ``upscale``.
"""
instance = node_class()
declared = getattr(node_class, "FUNCTION", None)
candidates = []
if isinstance(declared, str) and declared:
candidates.append(declared)
candidates.extend(("execute", "run", "process", "sample", "upscale"))
seen = set()
for name in candidates:
if name in seen:
continue
seen.add(name)
method = getattr(instance, name, None)
if callable(method):
return _unwrap_node_output(method(**kwargs))
raise RuntimeError(
"MMH3UltimateUpscale exposes no callable ComfyUI FUNCTION method "
f"(declared={declared!r})"
)
def _one(value: Any, name: str) -> Any:
"""Unwrap ComfyUI list inputs while rejecting ambiguous values."""
if isinstance(value, list):
if len(value) != 1:
raise ValueError(f"{name} must contain one shared value")
return value[0]
return value
def _clips(value: Any) -> list[Any]:
if isinstance(value, list):
return value
return [value]
def _per_clip(value: Any, count: int, name: str) -> list[Any]:
values = _clips(value)
if len(values) == 1:
return values * count
if len(values) != count:
raise ValueError(f"{name} has {len(values)} values for {count} clips")
return values
class H3ContinuumTiledLatentUpscalePerClip:
"""Upscale each Continuum physical clip separately with tiled H3 latent work.
This is intentionally a thin adapter around ComfyUI's native
``MMH3UltimateUpscale`` implementation. It does not vendor or replace
the model-specific implementation; it only supplies an independent
sampler/scheduler/noise/CFG schedule for each clip and preserves list
boundaries so clips never become one oversized VRAM job.
"""
DESCRIPTION = (
"Apply native MiniMax H3 tiled latent upscaling independently to every "
"Continuum physical clip. Clips remain separate LATENT list items."
)
SEARCH_ALIASES = [
"H3 tiled upscale per clip",
"MiniMax H3 clip upscale",
"Continuum latent upscale",
]
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"model": ("MODEL",),
"clip": ("CLIP",),
"video_latents": ("LATENT",),
"assembly_plan": ("H3_CONTINUUM_ASSEMBLY_PLAN",),
"refine_context": ("H3_CONTINUUM_REFINE_CONTEXT",),
"latent_upscale_param": ("H3_UPSCALE_PARAM",),
"spatial_split_param": ("H3_SPATIAL_PARAM",),
"upscale_sampler": (
_sampler_options(),
{"default": "euler_ancestral"},
),
"upscale_scheduler": (
_scheduler_options(),
{"default": "simple"},
),
"upscale_steps": ("INT", {"default": 3, "min": 1, "max": 50}),
"upscale_denoise": (
"FLOAT",
{"default": 0.2, "min": 0.0, "max": 1.0, "step": 0.01},
),
"upscale_cfg": (
"FLOAT",
{"default": 1.0, "min": 0.0, "max": 20.0, "step": 0.1},
),
"upscale_noise_seed": (
"INT",
{
"default": 363742558835919,
"min": 0,
"max": 0xFFFFFFFFFFFFFFFF,
"control_after_generate": True,
},
),
},
"optional": {
"conditioning": ("CONDITIONING",),
"video_vae": ("VAE",),
"negative": ("CONDITIONING",),
"temporal_split_param": ("H3_TEMPORAL_PARAM",),
"upscale_sigmas": ("SIGMAS",),
},
}
INPUT_IS_LIST = True
RETURN_TYPES = ("LATENT", "STRING")
RETURN_NAMES = ("upscaled_video_latents", "report")
OUTPUT_IS_LIST = (True, False)
FUNCTION = "upscale"
CATEGORY = "MiniMax H3/Continuum/Upscale"
def upscale(
self,
model,
clip,
video_latents,
assembly_plan,
refine_context,
latent_upscale_param,
spatial_split_param,
upscale_sampler,
upscale_scheduler,
upscale_steps,
upscale_denoise,
upscale_cfg,
upscale_noise_seed,
conditioning=None,
video_vae=None,
negative=None,
temporal_split_param=None,
upscale_sigmas=None,
):
native = _native_node("MMH3UltimateUpscale")
random_noise = _native_node("RandomNoise")
sampler_select = _native_node("KSamplerSelect")
scheduler = _native_node("BasicScheduler")
missing = [
name
for name, value in (
("MMH3UltimateUpscale", native),
("RandomNoise", random_noise),
("KSamplerSelect", sampler_select),
("BasicScheduler", scheduler),
)
if value is None
]
if missing:
raise RuntimeError("Missing native ComfyUI nodes: " + ", ".join(missing))
latents = _clips(video_latents)
if not latents:
raise ValueError("video_latents must contain at least one clip")
if conditioning is None:
# The Continuum sampler's refine context is a runtime contract, not
# a Core CONDITIONING value. Reuse Continuum's own bridge so each
# physical clip receives its captured prompt/reference conditioning
# and its context-aware model clone at the target geometry.
from .second_pass import prepare_physical_refine_groups
prepared = prepare_physical_refine_groups(
model=_one(model, "model"),
clip=_one(clip, "clip"),
video_latents=latents,
assembly_plan=_one(assembly_plan, "assembly_plan"),
refine_context=_one(refine_context, "refine_context"),
video_vae=(
None if video_vae is None else _one(video_vae, "video_vae")
),
)
models = prepared["group_models"]
conditionings = prepared["conditioning"]
if len(models) != len(latents) or len(conditionings) != len(latents):
raise ValueError(
"refine context bridge did not produce one model and conditioning "
"value per physical clip"
)
else:
models = _per_clip(model, len(latents), "model")
conditionings = _per_clip(conditioning, len(latents), "conditioning")
negatives = (
[None] * len(latents)
if negative is None
else _per_clip(negative, len(latents), "negative")
)
temporal = (
[None] * len(latents)
if temporal_split_param is None
else _per_clip(temporal_split_param, len(latents), "temporal_split_param")
)
sigma_values = (
[None] * len(latents)
if upscale_sigmas is None
else _per_clip(upscale_sigmas, len(latents), "upscale_sigmas")
)
sampler_name = str(_one(upscale_sampler, "upscale_sampler"))
scheduler_name = str(_one(upscale_scheduler, "upscale_scheduler"))
steps = int(_one(upscale_steps, "upscale_steps"))
denoise = float(_one(upscale_denoise, "upscale_denoise"))
cfg = float(_one(upscale_cfg, "upscale_cfg"))
seed = int(_one(upscale_noise_seed, "upscale_noise_seed"))
upscale_param = _one(latent_upscale_param, "latent_upscale_param")
spatial_param = _one(spatial_split_param, "spatial_split_param")
outputs = []
reports = []
for index, (clip_latent, clip_model, clip_conditioning) in enumerate(
zip(latents, models, conditionings), start=1
):
samples = clip_latent.get("samples") if isinstance(clip_latent, dict) else None
if samples is None or getattr(samples, "ndim", None) != 5:
raise ValueError(f"clip {index} is not a 5-D H3 LATENT")
noise = _unwrap_single_output(
random_noise().get_noise(seed + index - 1), "RandomNoise"
)
sampler = _unwrap_single_output(
sampler_select().get_sampler(sampler_name), "KSamplerSelect"
)
sigmas = sigma_values[index - 1]
if sigmas is None:
sigmas = _unwrap_single_output(
scheduler().get_sigmas(
clip_model, scheduler_name, steps, denoise
),
"BasicScheduler",
)
kwargs = {
"model": clip_model,
"conditioning": clip_conditioning,
"latent": clip_latent,
"noise": noise,
"sampler": sampler,
"sigmas": sigmas,
"cfg": cfg,
"latent_upscale_param": upscale_param,
"temporal_split_param": temporal[index - 1],
"spatial_split_param": spatial_param,
}
if negatives[index - 1] is not None:
kwargs["negative"] = negatives[index - 1]
result = _call_native(native, kwargs)
if not isinstance(result, (tuple, list)) or not result:
raise RuntimeError(f"native upscale returned no output for clip {index}")
output = result[0]
output_samples = output.get("samples") if isinstance(output, dict) else None
if output_samples is None or output_samples.ndim != 5:
raise RuntimeError(f"native upscale returned an invalid clip {index}")
if output_samples.shape[2] != samples.shape[2]:
raise RuntimeError(f"tiled upscale changed temporal length for clip {index}")
outputs.append(output)
reports.append(
f"clip {index}: {tuple(samples.shape)} -> {tuple(output_samples.shape)}"
)
return (
outputs,
"H3 Continuum per-clip tiled upscale; "
f"clips={len(outputs)}, sampler={sampler_name}, scheduler={scheduler_name}, "
f"steps={steps}, denoise={denoise:g}, cfg={cfg:g}, seed_base={seed}; "
+ "; ".join(reports),
)
NODE_CLASS_MAPPINGS = {
"H3ContinuumTiledLatentUpscalePerClip": H3ContinuumTiledLatentUpscalePerClip,
}
NODE_DISPLAY_NAME_MAPPINGS = {
"H3ContinuumTiledLatentUpscalePerClip": (
"H3 Continuum Tiled Latent Upscale (Per Clip)"
),
}
+1 -1
View File
@@ -1,6 +1,6 @@
"""Package version and public schema identifiers."""
PACKAGE_VERSION = "3.9.4"
PACKAGE_VERSION = "3.9.10"
STATE_SCHEMA_VERSION = 1
PLAN_SCHEMA_VERSION = 1
PUBLIC_API_VERSION = 3
+70 -8
View File
@@ -23,6 +23,7 @@ const LEGACY_RUN_NAME_WIDGET = "run_name";
const CHUNKS_WIDGET = "chunks";
const REGENERATE_WIDGET = "reroll_from_chunk";
const REROLL_NONCE_WIDGET = "reroll_nonce";
const REGENERATE_INTENT_PROPERTY = "h3_continuum_regenerate_from";
const RUN_STORAGE_WIDGET = "run_storage";
const REFERENCE_SIZE_WIDGET = "reference_size";
const TIMELINE_SIZE_WIDGET = "timeline_video_size";
@@ -464,9 +465,18 @@ function setExistingWidgetValue(widget, value) {
widget.callback?.(value);
}
function hasManualRegenerateIntent(node) {
const value = findWidget(node, REGENERATE_WIDGET)?.value;
return Boolean(node.__h3ContinuumManualRegenerateIntent
|| (value && value !== "Auto"
&& node.properties?.[REGENERATE_INTENT_PROPERTY] === value));
}
function setRegenerateProgrammatically(node, value) {
const widget = findWidget(node, REGENERATE_WIDGET);
if (!widget) return;
delete node.__h3ContinuumManualRegenerateIntent;
if (node.properties) delete node.properties[REGENERATE_INTENT_PROPERTY];
node.__h3ContinuumProgrammaticRegenerate = true;
try {
setExistingWidgetValue(widget, value);
@@ -484,13 +494,16 @@ function repairStaleReviewRegenerate(node, project) {
!revision
|| !["review_ready", "complete"].includes(revision.status)
|| activeReviewPrompt(node)
|| node.__h3ContinuumManualRegenerateIntent
|| hasManualRegenerateIntent(node)
|| node.__h3ContinuumRestartSelected
) {
return false;
}
const widget = findWidget(node, REGENERATE_WIDGET);
if (!widget || widget.value === "Auto" || widget.value === 0) return false;
// Only the legacy unowned Chunk 1 / nonce 0 residue is ambiguous.
// Later boundaries and fixed variations are explicit saved settings.
if (!["Chunk 1", 1].includes(widget?.value)
|| Number(findWidget(node, REROLL_NONCE_WIDGET)?.value ?? 0) !== 0) return false;
setRegenerateProgrammatically(node, "Auto");
setExistingWidgetValue(findWidget(node, REROLL_NONCE_WIDGET), 0);
return true;
@@ -1573,7 +1586,7 @@ function reviewStatus(node) {
const revision = canonicalStorageRevision(project);
if (!revision) return "Queue the workflow to create the first chunk.";
if (takeReadOnly(revision)) {
return "Saved history is read-only for Sampling v6\nStart from Chunk 1 with current settings. Existing Takes are kept.";
return takeReadOnlyExplanation(revision);
}
if (revision.status === "complete") {
const unit = revision.review_unit;
@@ -1657,6 +1670,51 @@ function takeReadOnly(item) {
return Number(item?.sampling_contract_version) !== 6 || item?.generation_reusable !== true;
}
const TAKE_REUSE_REASON_HELP = new Map([
["KJ/Core preview implementation is outside the audited source profiles",
"Bypass KJ Model Preview Override in a workflow copy, then test a new Run."],
["KJ preview VAE/Audio VAE or Tiny VAE is outside the supported configuration",
"Disconnect the preview node's optional VAE/Audio VAE and set Tiny VAE to none. Keep the generation VAEs connected."],
["Core approximate-VAE preview is outside the supported configuration",
"Set Core preview to Auto or Latent2RGB before starting a new Run."],
["KJ preview runtime code or state could not be verified",
"Check Detailed Report; bypass KJ Model Preview Override in a workflow copy to isolate it."],
...[
"MODEL wrapper/patch contract is not completely observable",
"SAMPLER function/options are not completely observable",
"CLIP/Qwen encoder contract is not completely observable",
"Video VAE contract is not completely observable",
"Reference Audio VAE contract is not completely observable",
"Driving Audio VAE contract is not completely observable",
"Saved generation conditions could not be fully verified; see Detailed Report",
].map((reason) => [reason, ""]),
]);
function takeReadOnlyExplanation(item) {
if (!takeReadOnly(item)) return "";
if (Number(item?.sampling_contract_version) === 5) {
return "Sampling v5 history is read-only for v6 generation\nStart a new Run from Chunk 1. Existing Takes are kept.";
}
const verifiedMetadata = Number(item?.sampling_contract_version) === 6
&& item?.generation_reusable === false;
const lines = [verifiedMetadata
? "Saved Sampling v6 Take is read-only: its generation conditions were not fully verified."
: "Saved Take reuse could not be verified: compatibility metadata is incomplete or unsupported."];
const reasons = Array.isArray(item?.resume_disabled_reasons)
? [...new Set(item.resume_disabled_reasons.slice(0, 8).filter(
(reason) => typeof reason === "string" && TAKE_REUSE_REASON_HELP.has(reason),
))]
: [];
lines.push(...reasons);
const help = [...new Set(reasons.map((reason) => TAKE_REUSE_REASON_HELP.get(reason)).filter(Boolean))];
if (help.length) {
lines.push(...help, "After correcting the configuration, start a new Run from Chunk 1. Existing Takes are kept.");
} else {
lines.push("Check Detailed Report and the installed/backend version before starting a new Run. Existing Takes are kept.");
}
return lines.join("\n");
}
function takeCatalog(node) {
const project = node.__h3ContinuumTakeProject;
if (project && takeCatalogCache.has(project)) return takeCatalogCache.get(project);
@@ -1820,7 +1878,7 @@ function takeAncestry(project, selected, catalog) {
function continueFromTakeSummary(project, selected, catalog) {
if (takeReadOnly(selected)) {
return "Read-only history: start a new Full Video or Review from Chunk 1; saved Takes remain available.";
return "Continue From Here: unavailable\n" + takeReadOnlyExplanation(selected);
}
const chain = takeAncestry(project, selected, catalog);
if (!chain.length) {
@@ -1879,9 +1937,6 @@ function takeStatus(node) {
`Compatible history: ${catalog.filter(item => !takeReadOnly(item)).length} Takes`,
continueFromTakeSummary(project, selected, catalog),
];
if (takeReadOnly(selected)) {
lines.push(`Sampling v${selected.sampling_contract_version ?? "unknown"}: read-only; cannot reuse in v6 generation.`);
}
if (hiddenCount) {
lines.push(
`${hiddenCount} incompatible or incomplete Take${hiddenCount === 1 ? "" : "s"} hidden`,
@@ -2885,8 +2940,13 @@ function configureConditionalWidgets(node) {
if (!node.__h3ContinuumProgrammaticRegenerate) {
if (value === "Auto" || value === 0 || value == null) {
delete node.__h3ContinuumManualRegenerateIntent;
if (node.properties) delete node.properties[REGENERATE_INTENT_PROPERTY];
} else {
node.__h3ContinuumManualRegenerateIntent = true;
if (isModernSamplerClass(node.comfyClass)) {
node.properties ||= {};
node.properties[REGENERATE_INTENT_PROPERTY] = value;
}
}
}
return result;
@@ -3083,6 +3143,7 @@ function normalizeReviewActionOnLoad(node) {
if (!isModernSamplerClass(node.comfyClass)) {
return false;
}
delete node.__h3ContinuumManualRegenerateIntent;
delete node[REVIEW_PENDING_ACTION];
delete node[REVIEW_UI_SELECTION];
let changed = false;
@@ -3145,7 +3206,8 @@ function buildReviewQueueInputs(node, apiInputs) {
if (
setup
&& from === "Chunk 1"
&& !node.__h3ContinuumManualRegenerateIntent
&& Number(nonce) === 0
&& !hasManualRegenerateIntent(node)
&& !node.__h3ContinuumRestartSelected
) {
from = "Auto";
+30 -9
View File
@@ -38,9 +38,15 @@ function linkedSamplers(node) {
function destination(node) {
const targets = linkedSamplers(node);
const unknown = (reason) => ({ count: MAX, known: false, reason });
if (!targets.length) return unknown("no_sampler");
if (targets.some((target) => target.inputs?.some((input) => input.name === "chunks" && input.link != null))) {
return unknown("external_chunks");
}
const counts = targets.map((target) => Number(getWidget(target, "chunks")?.value));
const known = counts.length && counts.every((n) => Number.isInteger(n) && n >= 1 && n <= MAX && n === counts[0]);
return { count: known ? counts[0] : MAX, known: Boolean(known) };
if (!counts.every((n) => Number.isInteger(n) && n >= 1 && n <= MAX)) return unknown("unavailable_count");
if (!counts.every((n) => n === counts[0])) return unknown("different_counts");
return { count: counts[0], known: true, reason: "direct_count" };
}
function invalidate(node) {
@@ -122,10 +128,10 @@ export function configureV39ReferenceImages(node) {
target.callback?.(target.value);
};
const refresh = () => {
const { count, known } = destination(node);
const { count, known, reason } = destination(node);
const rows = Array.from({ length: 9 }, (_, i) => i + 1).filter((slot) =>
node.inputs?.some((input) => input.name === `reference_image_${slot}` && input.link != null));
const key = JSON.stringify([mode.value, count, known, rows, selectors.map((item) => item.value), page]);
const key = JSON.stringify([mode.value, count, known, reason, rows, selectors.map((item) => item.value), page]);
if (key === previousKey) return;
previousKey = key;
panel.replaceChildren();
@@ -134,10 +140,20 @@ export function configureV39ReferenceImages(node) {
reservePanelSpace(54);
return;
}
panel.append(element("div", known ? `Per-chunk assignments · ${count} chunks` : "Destination unknown · show chunks 1–16"));
panel.append(element("div", known ? `Per-chunk assignments · ${count} chunks` : "Per-chunk assignments · range 1–16"));
if (!known) {
const explanation = {
no_sampler: "Connect this helper to a Sampler to determine its chunk count.",
external_chunks: "Chunk count is supplied by a connected node.",
unavailable_count: "The Sampler chunk count is not available.",
different_counts: "Connected Samplers have different chunk counts.",
};
panel.append(element("div", explanation[reason]));
panel.append(element("div", "Assignments apply only to chunks generated by the Sampler."));
}
if (!rows.length) {
panel.append(element("div", "Connect Reference Images to assign them."));
reservePanelSpace(64);
reservePanelSpace(known ? 64 : 104);
return;
}
page = Math.min(page, Math.ceil(count / 6) - 1);
@@ -182,7 +198,7 @@ export function configureV39ReferenceImages(node) {
}
const actions = element("td");
for (const [label, on] of [["All", true], ["Off", false]]) {
const button = element("button", label);
const button = element("button", known ? label : `${label} 1–16`);
button.type = "button";
button.disabled = selected === null;
button.onclick = () => {
@@ -204,14 +220,19 @@ export function configureV39ReferenceImages(node) {
panel.append(element("div", `Image ${slot}: invalid saved selection.`));
continue;
}
if (!known) {
const saved = [...selected].sort((a, b) => a - b);
panel.append(element("div", `Image ${slot} → Saved assignments: ${saved.join(", ") || "none"}`));
continue;
}
const active = [...selected].filter((n) => n <= count).sort((a, b) => a - b);
const future = [...selected].filter((n) => n > count).sort((a, b) => a - b);
panel.append(element("div", `Image ${slot} → Chunks ${active.join(", ") || "none"}${future.length ? ` · saved beyond range: ${future.join(", ")}` : ""}`));
}
if (rows.every((slot) => ![...(savedChunks(selectors[slot - 1].value) || [])].some((n) => n <= count))) {
panel.append(element("div", "No references assigned. Check the chunks to use."));
panel.append(element("div", known ? "No references assigned. Check the chunks to use." : "No saved assignments. Check the chunks to use."));
}
reservePanelSpace(Math.min(360, 77 + rows.length * 36 + (count > 6 ? 32 : 0)));
reservePanelSpace(Math.min(360, 77 + rows.length * 36 + (count > 6 ? 32 : 0) + (known ? 0 : 40)));
node.setDirtyCanvas?.(true, true);
};
node.__h3ReferenceImagesRefresh = refresh;
-656
View File
@@ -1,656 +0,0 @@
import { app } from "../../scripts/app.js";
const CLASS = "DumasH3SequencePromptBuilder";
const MAX_CLIPS = 16;
const FIXED_TEXT_WIDGETS = [
["shared_directive", 84],
["soundscape", 60],
["music", 60],
];
const DURATION_OPTIONS = [
["1–5", 5, 2, 0.21], ["6–22", 22, 7, 0.92], ["23–39", 39, 12, 1.63],
["40–56", 56, 17, 2.33], ["57–73", 73, 22, 3.04], ["74–90", 90, 27, 3.75],
["91–107", 107, 32, 4.46], ["108–124", 124, 37, 5.17], ["125–141", 141, 42, 5.88],
["142–158", 158, 47, 6.58], ["159–175", 175, 52, 7.29], ["176–192", 192, 57, 8.00],
["193–209", 209, 62, 8.71], ["210–226", 226, 67, 9.42], ["227–243", 243, 72, 10.13],
["244–260", 260, 77, 10.83], ["261–277", 277, 82, 11.54], ["278–294", 294, 87, 12.25],
["295–311", 311, 92, 12.96], ["312–328", 328, 97, 13.67], ["329–345", 345, 102, 14.38],
["346–362", 362, 107, 15.08],
].map(([requested, actual, latent, labelSeconds]) => ({
requested, actual, latent, labelSeconds, seconds: actual / 24,
}));
const CAMERA_HELPERS = [
["Push / Pull", ["The camera pushes in", "The camera pulls out"]],
["Zoom", ["The camera zooms in", "The camera zooms out"]],
["Pan", ["The camera pans left", "The camera pans right"]],
["Truck", ["The camera trucks left", "The camera trucks right"]],
["Tilt", ["The camera tilts up", "The camera tilts down"]],
["Pedestal", ["The camera pedestals up", "The camera pedestals down"]],
["Movement", ["The camera holds a static shot", "The camera arcs around the subject", "The camera tracks the subject"]],
["POV", ["The camera holds the subject's point of view"]],
["Roll", ["The camera rolls clockwise", "The camera rolls counterclockwise"]],
["Shake", ["The camera shakes slightly", "The camera shakes strongly"]],
];
const get = (node, name) => node.widgets?.find((widget) => widget.name === name);
const el = (tag, text = "") => { const item = document.createElement(tag); item.textContent = text; return item; };
function defaultState(count, seconds) {
return { version: 1, clips: Array.from({ length: count }, () => ({
duration: seconds, slots: [{ start: 0, end: seconds, text: "" }], prompt: "",
})) };
}
function readState(widget, count, seconds) {
try {
const value = JSON.parse(String(widget?.value || ""));
if (Array.isArray(value.clips)) return value;
} catch (_) { /* replace malformed editor state with a usable default */ }
return defaultState(count, seconds);
}
function hide(widget) {
if (!widget) return;
widget.type = "hidden"; widget.hidden = true; widget.options ||= {};
widget.options.hidden = true; widget.computeSize = () => [0, 0]; widget.draw = () => {};
}
function configure(node) {
if (node?.comfyClass !== CLASS || typeof document === "undefined" || typeof node.addDOMWidget !== "function") return;
if (node.__dumasSequencePromptBuilder) return node.__dumasSequencePromptBuilder();
const countWidget = get(node, "clip_count");
const secondsWidget = get(node, "default_chunk_seconds");
const stateWidget = get(node, "timeline_json");
if (!countWidget || !secondsWidget || !stateWidget) return;
hide(secondsWidget);
hide(stateWidget);
FIXED_TEXT_WIDGETS.forEach(([name, height]) => {
const widget = get(node, name);
if (!widget) return;
widget.options ||= {};
widget.options.multiline = true;
widget.__dumasFixedHeight = height;
widget.computeSize = (width) => [
Math.max(300, Number(width) || Number(node.size?.[0]) || 420),
height,
];
});
const panel = el("div");
panel.style.cssText = "box-sizing:border-box;position:relative;width:calc(100% + 14px);margin-left:-7px;min-width:420px;padding:9px;color:#d9e3dc;background:#292929;border:1px solid #454545;border-radius:4px;font:12px Arial;pointer-events:auto";
const host = el("div"); host.style.cssText = "box-sizing:border-box;pointer-events:auto;background:#292929;width:100%"; host.append(panel);
["pointerdown", "mousedown", "mouseup", "dblclick"].forEach((name) => host.addEventListener(name, (event) => event.stopPropagation()));
const display = node.addDOMWidget("clip_tabs", "dumas_h3_clip_tabs", host, { serialize: false });
if (!node.__dumasSequenceResizeWrapped) {
const previousResize = node.onResize;
node.onResize = function (size) {
if (this.__dumasSequenceMinHeight > 0 && size[1] < this.__dumasSequenceMinHeight) {
size[1] = this.__dumasSequenceMinHeight;
}
return previousResize?.apply(this, arguments);
};
node.__dumasSequenceResizeWrapped = true;
}
display.options ||= {}; display.options.serialize = false;
let active = 0;
let state = readState(stateWidget, Number(countWidget.value) || 2, Number(secondsWidget.value) || 4);
let promptText = null;
// Writing the hidden widget value is safe while a textarea owns focus.
// Calling its callback is not: ComfyUI callbacks redraw the DOM widget,
// which destroys the focused textarea during blur/pointer transitions.
const persistState = () => {
stateWidget.value = JSON.stringify(state);
node.setDirtyCanvas?.(true, true);
};
const setState = () => {
persistState();
stateWidget.callback?.(stateWidget.value);
};
const input = (type, value, onChange) => {
const item = el("input"); item.type = type; item.value = value;
item.style.cssText = type === "range"
? "width:100%;box-sizing:border-box;margin:2px 0;accent-color:#79c98a"
: "width:100%;box-sizing:border-box;margin:2px 0;color:#111";
item.onchange = () => onChange(item.value); return item;
};
const addSlot = (clip, afterIndex = clip.slots.length - 1) => {
const duration = Number(clip.duration) || 0.1;
if (!clip.slots.length) {
clip.slots.push({ start: 0, end: duration, text: "" });
return;
}
const newCount = clip.slots.length + 1;
// Keep a minimum only when the selected duration allows it. The old
// fixed 0.1s floor made later shots collapse at short frame choices.
const minSlice = Math.min(0.1, duration / newCount);
let splitIndex = Math.max(0, Math.min(afterIndex, clip.slots.length - 1));
const lengthOf = (slot) => Math.max(0, Number(slot.end) - Number(slot.start));
if (lengthOf(clip.slots[splitIndex]) < minSlice * 2) {
splitIndex = clip.slots.reduce(
(best, slot, index) => lengthOf(slot) > lengthOf(clip.slots[best]) ? index : best,
0,
);
}
const target = clip.slots[splitIndex];
const start = Number(target?.start) || 0;
const end = Math.min(duration, Number(target?.end) || duration);
let split = start + (end - start) / 2;
split = Math.max(start + minSlice, Math.min(end - minSlice, split));
if (!(split > start && split < end)) {
// This is only reachable for malformed or extremely short state;
// redistribute boundaries evenly so the new shot is still real.
const texts = clip.slots.map((slot) => slot.text || "");
texts.splice(splitIndex + 1, 0, "");
clip.slots = texts.map((text, index) => ({
start: duration * index / newCount,
end: duration * (index + 1) / newCount,
text,
}));
return;
}
target.end = split;
clip.slots.splice(splitIndex + 1, 0, { start: split, end, text: "" });
};
const formatTime = (seconds) => `00:${Number(seconds).toFixed(3).padStart(6, "0")}`;
// Keep the user's script as the source of truth, like CSGlide does. The
// slots are only the timeline model used to paint the bar and to locate
// boundaries. Rebuilding the textarea from slots is lossy (and was the
// reason a redraw could erase text that had just been typed).
const formatPromptFromSlots = (clip) => clip.slots.map((slot, index) =>
`[Shot ${index + 1}] At ${formatTime(slot.start)}, ${slot.text || ""}`).join("\n\n");
const promptFor = (clip) => {
if (typeof clip.prompt === "string" && clip.prompt.length) return clip.prompt;
return formatPromptFromSlots(clip);
};
const parseTime = (value) => {
const parts = String(value || "").split(":").map(Number);
if (parts.some((part) => !Number.isFinite(part))) return NaN;
if (parts.length === 3) return parts[0] * 3600 + parts[1] * 60 + parts[2];
if (parts.length === 2) return parts[0] * 60 + parts[1];
return parts[0];
};
const parsePromptSlots = (clip, value) => {
const text = String(value || "");
const markerPattern = /\[Shot\s+\d+\]\s+At\s+((?:\d+(?::\d+){1,2}(?:\.\d+)?|\d+(?:\.\d+)?))\s*,\s*/gi;
const markers = [...text.matchAll(markerPattern)];
if (!markers.length) return false;
const duration = Math.max(0.1, Number(clip.duration) || 0.1);
const slots = markers.map((marker, index) => {
const rawStart = parseTime(marker[1]);
const start = Number.isFinite(rawStart)
? Math.min(Math.max(0, rawStart), Math.max(0, duration - 0.1))
: 0;
const next = markers[index + 1];
const rawEnd = next ? parseTime(next[1]) : duration;
const end = Math.min(duration, Math.max(start + 0.1, Number.isFinite(rawEnd) ? rawEnd : duration));
const from = marker.index + marker[0].length;
const to = next ? next.index : text.length;
return { start, end, text: text.slice(from, to).trim() };
});
slots.forEach((slot, index) => {
if (index > 0) slot.start = slots[index - 1].end;
if (slot.end <= slot.start) slot.end = Math.min(duration, slot.start + 0.1);
});
slots[slots.length - 1].end = duration;
clip.slots = slots;
return true;
};
const syncPrompt = (clip, value) => {
const text = String(value || "");
if (parsePromptSlots(clip, text)) return true;
// The shot description starts on the same line as the marker in the
// CSGlide-style editor. Only consume the marker prefix through its
// timestamp comma; consuming the whole line would discard text such
// as `[Shot 1] At 00:00.000, the character walks...` when a new shot
// is added and the marker scaffold is rebuilt.
const marks = [...text.matchAll(/\[Shot\s+(\d+)\]\s+At\s+[^,\n]+,\s*/gi)];
if (!marks.length) { if (clip.slots[0]) clip.slots[0].text = text.trim(); return; }
marks.forEach((mark, index) => {
const slotIndex = Number(mark[1]) - 1;
if (!clip.slots[slotIndex]) return;
const from = mark.index + mark[0].length;
const to = index + 1 < marks.length ? marks[index + 1].index : text.length;
clip.slots[slotIndex].text = text.slice(from, to).trim();
});
return false;
};
const retimePrompt = (text, clip) => {
let index = 0;
return String(text || "").replace(/(\[Shot\s+\d+\]\s+At\s+)[^,\n]+(,?)/gi, (whole, prefix, suffix) => {
const start = clip.slots[index++]?.start ?? 0;
return `${prefix}${formatTime(start)}${suffix}`;
});
};
const commitPrompt = (clip, value, redraw = false) => {
clip.prompt = String(value ?? "");
syncPrompt(clip, clip.prompt);
// Do not invoke the widget callback for ordinary text edits. The
// callback redraws the DOM and can run while the textarea is blurring
// to a timeline handle, exactly the interaction CSGlide avoids.
persistState();
if (redraw) render();
};
const rebuildPromptAfterStructureChange = (clip) => {
// The caller syncs the current text before changing the slot list.
// Do not parse clip.prompt again here: after adding a shot it still
// has the old marker count, and reparsing would discard the new slot.
clip.prompt = formatPromptFromSlots(clip);
};
const render = () => {
const count = Math.max(1, Math.min(MAX_CLIPS, Number(countWidget.value) || 1));
const currentSeconds = Number(secondsWidget.value) || DURATION_OPTIONS[6].seconds;
const selectedDuration = DURATION_OPTIONS.reduce((best, option) =>
Math.abs(option.seconds - currentSeconds) < Math.abs(best.seconds - currentSeconds) ? option : best,
DURATION_OPTIONS[6]);
if (Number(secondsWidget.value) !== selectedDuration.seconds) secondsWidget.value = selectedDuration.seconds;
const chunkDuration = selectedDuration.seconds;
while (state.clips.length < count) state.clips.push({ duration: chunkDuration, slots: [] });
state.clips = state.clips.slice(0, count); active = Math.min(active, count - 1);
panel.replaceChildren();
const tabs = el("div"); tabs.style.cssText = "display:flex;flex-wrap:wrap;gap:4px;margin-bottom:8px;border-bottom:1px solid #444;padding-bottom:6px";
state.clips.forEach((clip, index) => {
const button = el("button", `Clip ${index + 1}`); button.type = "button"; button.disabled = index === active;
button.style.cssText = `padding:3px 10px;border:1px solid ${index === active ? "#8bc99b" : "#555"};border-radius:3px;background:${index === active ? "#4b6f55" : "#333"};color:${index === active ? "#fff" : "#bbb"};cursor:pointer`;
button.onclick = () => { active = index; render(); }; tabs.append(button);
});
panel.append(tabs);
const clip = state.clips[active];
// All clips share the one authoritative project duration widget.
const duration = chunkDuration;
clip.duration = duration;
if (clip.slots.length) clip.slots[clip.slots.length - 1].end = duration;
promptText = promptFor(clip);
const durationControl = el("div");
durationControl.style.cssText = "display:flex;align-items:center;gap:8px;margin:0 0 8px;color:#b7d6bf";
const durationLabel = el("span", "Chunk duration / frame choice");
durationLabel.style.cssText = "font-weight:bold;white-space:nowrap";
const durationSelect = document.createElement("select");
durationSelect.style.cssText = "flex:1;min-width:0;box-sizing:border-box;padding:4px 6px;color:#e8e8e8;background:#171717;border:1px solid #596b60;border-radius:4px;font:11px monospace";
DURATION_OPTIONS.forEach((option) => {
const item = document.createElement("option");
item.value = String(option.seconds);
item.textContent = `${option.actual} Real Frames - ${option.latent} Latent Frames - ${option.labelSeconds.toFixed(2)}s`;
item.selected = option === selectedDuration;
durationSelect.append(item);
});
durationSelect.title = "Choose a native H3 frame length; actual frames are encoded from the latent-frame choice.";
durationSelect.onchange = () => {
secondsWidget.value = Number(durationSelect.value);
secondsWidget.callback?.(secondsWidget.value);
};
durationControl.append(durationLabel, durationSelect);
panel.append(durationControl);
const durationRow = el("div"); durationRow.style.cssText = "display:flex;align-items:center;gap:8px;margin:4px 0 6px;color:#b7d6bf";
durationRow.append(el("strong", `Clip ${active + 1}`), el("span", `· ${selectedDuration.actual} actual frames · ${selectedDuration.latent} latent frames · ${selectedDuration.labelSeconds.toFixed(2)}s shared chunk duration`)); panel.append(durationRow);
const timeline = el("div"); timeline.style.cssText = "position:relative;height:38px;margin:8px 0 4px;background:#171717;border-radius:4px;overflow:hidden;border:1px solid #555";
const palette = ["#4b9b67", "#4d7fb1", "#a87945", "#8c5aa8", "#a44f5d", "#3f8e8a"];
let prompt = null;
let promptSelectionStart = 0;
let promptSelectionEnd = 0;
let autocomplete = null;
let autocompleteRange = null;
let promptFrame = null;
let suggestionLayer = null;
let encoder = null;
const focusPrompt = () => {
if (!prompt) return;
try {
prompt.focus({ preventScroll: true });
} catch (_) {
prompt.focus();
}
};
const segments = [];
const handles = [];
const updateTimeline = () => {
clip.slots.forEach((slot, index) => {
const left = Math.max(0, Number(slot.start) / duration * 100);
const width = Math.max(0.1, (Number(slot.end) - Number(slot.start)) / duration * 100);
segments[index].style.left = `${left}%`;
segments[index].style.width = `${width}%`;
segments[index].title = `${Number(slot.start).toFixed(1)}s–${Number(slot.end).toFixed(1)}s`;
});
handles.forEach(({ handle, slot }) => { handle.style.left = `${Number(slot.end) / duration * 100}%`; });
};
clip.slots.forEach((slot, index) => {
const segment = el("button", `Shot ${index + 1}\n${(slot.text || "unscripted").slice(0, 18)}`); segment.type = "button";
segment.style.cssText = `position:absolute;top:0;height:100%;border:0;border-right:2px solid #292929;color:white;background:${palette[index % palette.length]};white-space:pre-wrap;font-size:10px;overflow:hidden;padding:2px;cursor:pointer`;
segment.onclick = (event) => {
event.preventDefault(); event.stopPropagation();
focusPrompt();
}; segments.push(segment); timeline.append(segment);
});
clip.slots.slice(0, -1).forEach((slot, index) => {
const next = clip.slots[index + 1];
const handle = el("div"); handle.title = `Resize boundary between Shot ${index + 1} and Shot ${index + 2}`;
handle.style.cssText = "position:absolute;top:0;width:17px;height:100%;transform:translateX(-50%);background:transparent;border:0;border-radius:3px;cursor:ew-resize;z-index:5;box-sizing:border-box;touch-action:none";
const ink = el("div");
ink.style.cssText = "position:absolute;left:7px;top:0;width:2px;height:100%;background:#f1f7f2;border:0;border-radius:1px;pointer-events:none";
handle.append(ink);
const move = (event) => {
event.preventDefault(); event.stopPropagation();
const rect = timeline.getBoundingClientRect();
const point = Math.min(duration - 0.1, Math.max(0.1, ((event.clientX - rect.left) / rect.width) * duration));
slot.end = point; next.start = point; updateTimeline();
};
let dragging = false;
const finish = (event) => {
if (!dragging) return;
dragging = false;
event?.preventDefault?.(); event?.stopPropagation?.();
document.removeEventListener("pointermove", move, true);
window.removeEventListener("pointerup", finish, true);
document.removeEventListener("mousemove", moveMouse, true);
document.removeEventListener("mouseup", finish, true);
promptText = retimePrompt(promptText ?? prompt?.value ?? clip.prompt ?? "", clip);
commitPrompt(clip, promptText);
// The timeline is already updated in place. Avoid rebuilding
// the panel, which moves the textarea and canvas viewport.
if (prompt) prompt.value = promptText;
if (encoder) encoder.textContent = clip.slots
.map((item, itemIndex) => `Shot ${itemIndex + 1} ${formatTime(item.start)}`)
.join(" ");
};
const moveMouse = (event) => { if (dragging) move(event); };
handle.onpointerdown = (event) => {
event.preventDefault();
event.stopPropagation();
dragging = true;
// Textarea edits are intentionally committed on change, but a
// drag can happen while it still has focus. Capture them before
// the timeline redraws the prompt from state.
if (prompt) { promptText = prompt.value; clip.prompt = promptText; }
handle.setPointerCapture?.(event.pointerId);
document.addEventListener("pointermove", move, true);
window.addEventListener("pointerup", finish, true);
document.addEventListener("mousemove", moveMouse, true);
document.addEventListener("mouseup", finish, true);
};
handles.push({ handle, slot }); timeline.append(handle);
});
updateTimeline();
panel.append(timeline);
const sliderHint = el("div", "Drag the white handles to resize shots. The bar represents the complete chunk duration."); sliderHint.style.cssText = "color:#aab6ad;margin-bottom:6px;font-size:11px"; panel.append(sliderHint);
const shotChips = el("div"); shotChips.style.cssText = "display:flex;gap:4px;overflow:hidden;margin:3px 0 5px";
clip.slots.forEach((slot, index) => {
const chipText = (slot.text || `Shot ${index + 1}`).replace(/\s+/g, " ").trim().slice(0, 31);
const chip = el("button", chipText || `Shot ${index + 1}`); chip.type = "button";
chip.title = `Jump to Shot ${index + 1}`;
chip.style.cssText = `max-width:180px;overflow:hidden;text-overflow:ellipsis;white-space:nowrap;padding:3px 8px;border:1px solid ${palette[index % palette.length]};border-radius:4px;background:#1e1e1e;color:#e6e6e6;font:11px Arial;cursor:pointer`;
chip.onclick = (event) => {
event.preventDefault(); event.stopPropagation();
if (!prompt) return;
const marker = `[Shot ${index + 1}]`;
const position = prompt.value.indexOf(marker);
focusPrompt();
if (position >= 0) prompt.setSelectionRange(position, position + marker.length);
};
shotChips.append(chip);
});
panel.append(shotChips);
const referenceRow = el("div"); referenceRow.style.cssText = "display:flex;gap:8px;overflow:hidden;margin:7px 0 8px;align-items:center;flex-wrap:wrap";
const connectedReferences = Array.from({ length: 9 }, (_, offset) => offset + 1)
.filter((index) => node.inputs?.some((input) => input.name === `reference_image_${index}` && input.link != null));
if (connectedReferences.length) {
connectedReferences.forEach((index) => {
const button = el("button"); button.type = "button";
button.title = `Insert reference image ${index} as @image${index}`;
button.style.cssText = "display:inline-flex;align-items:center;gap:7px;height:31px;padding:2px 11px 2px 3px;border:2px solid #49c9f5;border-radius:18px;background:#202b31;color:#70d8ff;font:16px monospace;cursor:pointer;white-space:nowrap";
const thumbnail = document.createElement("img");
thumbnail.alt = ""; thumbnail.draggable = false;
thumbnail.style.cssText = "width:25px;height:25px;object-fit:cover;border-radius:50%;background:#68727a;display:block";
const linkId = node.inputs?.find((input) => input.name === `reference_image_${index}`)?.link;
const link = linkId == null ? null : (app.graph?.links?.get?.(linkId) ?? app.graph?.links?.[linkId]);
const source = link?.origin_id == null ? null : app.graph?.getNodeById?.(link.origin_id);
const imageWidget = source?.widgets?.find((widget) => ["image", "filename", "file"].includes(widget.name));
const imageValue = imageWidget?.value;
if (imageValue && typeof imageValue === "object" && imageValue.filename) {
const params = new URLSearchParams({ filename: imageValue.filename, type: imageValue.type || "input", subfolder: imageValue.subfolder || "" });
thumbnail.src = `/view?${params.toString()}`;
} else if (typeof imageValue === "string" && imageValue) {
const params = new URLSearchParams({ filename: imageValue, type: "input", subfolder: "" });
thumbnail.src = `/view?${params.toString()}`;
} else {
thumbnail.src = "data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' width='25' height='25' viewBox='0 0 25 25'%3E%3Crect width='25' height='25' fill='%2368727a'/%3E%3Cpath d='M4 19l5-6 3 3 3-4 6 7' fill='none' stroke='%23dce8ed' stroke-width='2'/%3E%3Ccircle cx='17' cy='8' r='2' fill='%23dce8ed'/%3E%3C/svg%3E";
}
thumbnail.onerror = () => { thumbnail.style.opacity = "0.35"; };
button.append(thumbnail, el("span", `@image${index}`));
const rememberPromptSelection = () => {
if (!prompt) return;
promptSelectionStart = prompt.selectionStart ?? prompt.value.length;
promptSelectionEnd = prompt.selectionEnd ?? promptSelectionStart;
};
button.onpointerdown = (event) => {
rememberPromptSelection();
event.preventDefault();
event.stopPropagation();
};
button.onmousedown = (event) => {
event.preventDefault();
event.stopPropagation();
};
button.onclick = (event) => {
event.preventDefault(); event.stopPropagation();
if (!prompt) return;
const start = Math.min(promptSelectionStart, prompt.value.length);
const end = Math.min(Math.max(start, promptSelectionEnd), prompt.value.length);
const token = `@R${index} `;
prompt.value = prompt.value.slice(0, start) + token + prompt.value.slice(end);
const caret = start + token.length;
focusPrompt(); prompt.setSelectionRange(caret, caret);
promptSelectionStart = caret; promptSelectionEnd = caret;
promptText = prompt.value; clip.prompt = prompt.value;
syncPrompt(clip, clip.prompt); persistState();
};
referenceRow.append(button);
});
}
panel.append(referenceRow);
const promptLabel = el("div", "PROMPT SCRIPT"); promptLabel.style.cssText = "color:#8fa898;font-size:10px;letter-spacing:.08em;margin:4px 0 3px"; panel.append(promptLabel);
prompt = el("textarea"); prompt.rows = 10; prompt.value = promptText; prompt.placeholder = "[Shot 1] At 00:00.000, describe the opening action...\n\nType @ at the start of a new line to add the next shot."; prompt.style.cssText = "width:100%;box-sizing:border-box;color:#e8e8e8;background:#171717;border:1px solid #4e4e4e;border-radius:4px;padding:8px;font:12px/1.45 monospace;margin-top:2px;resize:vertical;outline:none";
const rememberPromptSelection = () => {
promptSelectionStart = prompt.selectionStart ?? prompt.value.length;
promptSelectionEnd = prompt.selectionEnd ?? promptSelectionStart;
};
const closeAutocomplete = () => {
autocomplete?.remove();
autocomplete = null;
autocompleteRange = null;
};
const sourceThumbnail = (index) => {
const thumbnail = document.createElement("img");
thumbnail.alt = ""; thumbnail.draggable = false;
thumbnail.style.cssText = "width:24px;height:24px;object-fit:cover;border-radius:50%;background:#68727a;display:block;flex:none";
const linkId = node.inputs?.find((input) => input.name === `reference_image_${index}`)?.link;
const link = linkId == null ? null : (app.graph?.links?.get?.(linkId) ?? app.graph?.links?.[linkId]);
const source = link?.origin_id == null ? null : app.graph?.getNodeById?.(link.origin_id);
const imageWidget = source?.widgets?.find((widget) => ["image", "filename", "file"].includes(widget.name));
const imageValue = imageWidget?.value;
if (imageValue && typeof imageValue === "object" && imageValue.filename) {
const params = new URLSearchParams({ filename: imageValue.filename, type: imageValue.type || "input", subfolder: imageValue.subfolder || "" });
thumbnail.src = `/view?${params.toString()}`;
} else if (typeof imageValue === "string" && imageValue) {
const params = new URLSearchParams({ filename: imageValue, type: "input", subfolder: "" });
thumbnail.src = `/view?${params.toString()}`;
}
thumbnail.onerror = () => { thumbnail.style.opacity = "0.35"; };
return thumbnail;
};
const openAutocomplete = (at) => {
closeAutocomplete();
autocompleteRange = { start: at, end: at + 1 };
const menu = el("div");
menu.style.cssText = "display:block;visibility:visible;opacity:1;position:absolute;z-index:100;left:4px;min-width:220px;max-width:310px;padding:5px;background:#25282b;border:1px solid #4b5960;border-radius:8px;box-shadow:0 8px 24px #000b;color:#e7edf0;pointer-events:auto";
menu.style.top = "0px";
const connected = Array.from({ length: 9 }, (_, offset) => offset + 1)
.filter((index) => node.inputs?.some((input) => input.name === `reference_image_${index}` && input.link != null));
const addOption = (content, title, action) => {
const option = el("button"); option.type = "button"; option.title = title;
option.style.cssText = "display:flex;align-items:center;gap:9px;width:100%;padding:6px 8px;border:0;border-radius:5px;background:transparent;color:#e7edf0;text-align:left;font:12px monospace;cursor:pointer";
option.onpointerdown = (event) => { event.preventDefault(); event.stopPropagation(); };
option.onmouseenter = () => { option.style.background = "#3b5661"; };
option.onmouseleave = () => { option.style.background = "transparent"; };
option.onclick = (event) => { event.preventDefault(); event.stopPropagation(); action(); };
option.append(...content); menu.append(option);
};
connected.forEach((index) => addOption([
sourceThumbnail(index),
el("span", `@image${index}`),
(() => { const hint = el("span", `<Picture ${index}>`); hint.style.cssText = "margin-left:auto;color:#84939a;font-size:10px"; return hint; })(),
], `Insert @R${index}`, () => {
const start = autocompleteRange?.start ?? prompt.value.length;
const end = autocompleteRange?.end ?? start;
const token = `@R${index} `;
prompt.value = prompt.value.slice(0, start) + token + prompt.value.slice(end);
const caret = start + token.length;
focusPrompt(); prompt.setSelectionRange(caret, caret);
promptSelectionStart = caret; promptSelectionEnd = caret;
promptText = prompt.value; clip.prompt = prompt.value;
syncPrompt(clip, clip.prompt); persistState(); closeAutocomplete();
}));
addOption([
(() => { const icon = el("span", "S"); icon.style.cssText = "display:grid;place-items:center;width:24px;height:24px;border-radius:50%;background:#397a91;color:#dff7ff;font-weight:bold"; return icon; })(),
el("span", "[Shot " + (clip.slots.length + 1) + "]"),
(() => { const hint = el("span", "new shot"); hint.style.cssText = "margin-left:auto;color:#84939a;font-size:10px"; return hint; })(),
], "Create the next timed shot", () => {
const start = autocompleteRange?.start ?? prompt.value.length;
prompt.value = prompt.value.slice(0, start) + prompt.value.slice(start + 1);
promptText = prompt.value; clip.prompt = prompt.value;
syncPrompt(clip, clip.prompt); addSlot(clip); rebuildPromptAfterStructureChange(clip);
closeAutocomplete(); setState(); render();
});
panel.append(menu);
menu.style.left = `${Math.max(5, (suggestionLayer?.offsetLeft || 0) + 4)}px`;
menu.style.top = `${Math.max(5, suggestionLayer?.offsetTop || 0)}px`;
autocomplete = menu;
};
prompt.oninput = () => { rememberPromptSelection(); promptText = prompt.value; clip.prompt = prompt.value; };
prompt.onpaste = () => {
// Let the browser insert the pasted text first, then parse its
// shot markers and redraw the track without losing the paste.
setTimeout(() => {
const changed = syncPrompt(clip, prompt.value);
promptText = prompt.value;
clip.prompt = prompt.value;
persistState();
if (changed) render();
}, 0);
};
prompt.onselect = rememberPromptSelection;
prompt.onkeyup = rememberPromptSelection;
prompt.onmouseup = rememberPromptSelection;
prompt.onblur = () => { rememberPromptSelection(); clip.prompt = prompt.value; syncPrompt(clip, clip.prompt); persistState(); };
prompt.onchange = () => { commitPrompt(clip, prompt.value); };
prompt.onpointerdown = (event) => event.stopPropagation();
prompt.onkeydown = (event) => {
if (event.key !== "@") return;
event.preventDefault();
const start = prompt.selectionStart ?? prompt.value.length;
const end = prompt.selectionEnd ?? start;
prompt.value = prompt.value.slice(0, start) + "@" + prompt.value.slice(end);
prompt.setSelectionRange(start + 1, start + 1);
promptSelectionStart = start;
promptSelectionEnd = start + 1;
promptText = prompt.value; clip.prompt = prompt.value;
syncPrompt(clip, clip.prompt); persistState();
openAutocomplete(start);
};
promptFrame = el("div");
promptFrame.style.cssText = "position:relative;overflow:visible;width:100%;box-sizing:border-box";
suggestionLayer = el("div");
suggestionLayer.style.cssText = "position:relative;z-index:100;height:0;width:100%;pointer-events:none;overflow:visible";
promptFrame.append(prompt);
panel.append(suggestionLayer);
panel.append(promptFrame);
promptSelectionStart = prompt.value.length;
promptSelectionEnd = prompt.value.length;
const insertPromptText = (text) => {
if (!prompt) return;
const start = Math.min(promptSelectionStart, prompt.value.length);
const end = Math.min(Math.max(start, promptSelectionEnd), prompt.value.length);
const before = prompt.value.slice(0, start);
const after = prompt.value.slice(end);
const prefix = before && !/\s$/.test(before) ? " " : "";
const token = `${prefix}${text} `;
prompt.value = before + token + after;
const caret = before.length + token.length;
focusPrompt(); prompt.setSelectionRange(caret, caret);
promptSelectionStart = caret; promptSelectionEnd = caret;
promptText = prompt.value; clip.prompt = prompt.value;
syncPrompt(clip, clip.prompt); persistState();
};
const helperPanel = el("div");
helperPanel.style.cssText = "display:flex;flex-wrap:wrap;gap:4px;align-items:center;margin:5px 0 7px;padding:5px 6px;background:#202020;border:1px solid #414141;border-radius:4px";
const helperTitle = el("span", "CAMERA HELPERS");
helperTitle.style.cssText = "width:100%;color:#8fa898;font-size:10px;letter-spacing:.08em;margin-bottom:1px";
helperPanel.append(helperTitle);
CAMERA_HELPERS.forEach(([category, phrases]) => phrases.forEach((phrase) => {
const button = el("button", phrase.replace(/^The camera /, ""));
button.type = "button";
button.title = `${category}: insert \"${phrase}\"`;
button.style.cssText = "padding:3px 7px;border:1px solid #596b60;border-radius:4px;background:#2b3b31;color:#b9e6c4;font:11px Arial;cursor:pointer;white-space:nowrap";
button.onpointerdown = (event) => { event.preventDefault(); event.stopPropagation(); };
button.onmousedown = (event) => { event.preventDefault(); event.stopPropagation(); };
button.onclick = (event) => { event.preventDefault(); event.stopPropagation(); insertPromptText(phrase); };
helperPanel.append(button);
}));
panel.insertBefore(helperPanel, promptFrame);
const checkLabel = el("div", "PROMPT CHECK"); checkLabel.style.cssText = "color:#8fa898;font-size:10px;letter-spacing:.08em;margin:7px 0 3px"; panel.append(checkLabel);
const check = el("div", `${clip.slots.length} shot${clip.slots.length === 1 ? "" : "s"} · ${duration.toFixed(1)}s chunk · @ references preserved`); check.style.cssText = "padding:6px 8px;background:#171717;border:1px solid #444;border-radius:4px;color:#9bd0a8;font:11px monospace"; panel.append(check);
const encoderLabel = el("div", "SENT TO ENCODER AS"); encoderLabel.style.cssText = "color:#8fa898;font-size:10px;letter-spacing:.08em;margin:7px 0 3px"; panel.append(encoderLabel);
encoder = el("div", clip.slots.map((slot, index) => `Shot ${index + 1} ${formatTime(slot.start)}`).join(" ")); encoder.style.cssText = "padding:6px 8px;background:#171717;border:1px solid #444;border-radius:4px;color:#7cc4d5;font:11px monospace;white-space:nowrap;overflow:hidden;text-overflow:ellipsis"; panel.append(encoder);
const actions = el("div"); actions.style.cssText = "display:flex;gap:5px;margin-top:5px";
const add = el("button", "@ Add shot"); add.type = "button"; add.onclick = () => { clip.prompt = prompt.value; syncPrompt(clip, clip.prompt); addSlot(clip); rebuildPromptAfterStructureChange(clip); setState(); render(); }; actions.append(add);
if (clip.slots.length > 1) { const remove = el("button", "Remove last shot"); remove.type = "button"; remove.onclick = () => { clip.prompt = prompt.value; syncPrompt(clip, clip.prompt); clip.slots.pop(); clip.slots.at(-1).end = clip.duration; rebuildPromptAfterStructureChange(clip); setState(); render(); }; actions.append(remove); }
panel.append(actions);
// The editor now contains chips, the prompt check and encoder summary;
// the old slot-count estimate left a large blank tail in the node.
// Measure the actual DOM panel instead of guessing from slot count.
const measuredHeight = Math.ceil(panel.getBoundingClientRect().height || panel.scrollHeight || 0);
display.computeSize = (width) => [Math.max(420, Number(width) || 420), measuredHeight + 2];
// CSGlide keeps the DOM panel floating but stores a node-level floor
// from measured content. Temporarily hide this widget from the
// ordinary LiteGraph calculation to get the non-panel chrome, then
// add the real panel height and clamp future manual resizes.
const realComputeSize = display.computeSize;
display.computeSize = () => [420, 0];
const chromeHeight = Number(node.computeSize?.()[1]) || 0;
display.computeSize = realComputeSize;
const width = Math.max(420, Number(node.size?.[0]) || 420);
const desiredHeight = chromeHeight + measuredHeight + 2;
node.__dumasSequenceMinHeight = desiredHeight;
if (typeof node.setSize === "function"
&& (Number(node.size?.[0]) !== width || Number(node.size?.[1]) < desiredHeight)) {
node.setSize([width, desiredHeight]);
}
};
const refresh = () => { state = readState(stateWidget, Number(countWidget.value) || 2, Number(secondsWidget.value) || 4); render(); setState(); };
countWidget.callback = ((previous) => function(...args) { const result = previous?.apply(this, args); refresh(); return result; })(countWidget.callback);
secondsWidget.callback = ((previous) => function(...args) {
const result = previous?.apply(this, args);
const duration = Math.max(0.1, Number(secondsWidget.value) || 0.1);
state.clips.forEach((clip) => {
clip.duration = duration;
clip.slots.forEach((slot) => {
slot.start = Math.min(Number(slot.start) || 0, Math.max(0, duration - 0.1));
slot.end = Math.min(Math.max(Number(slot.end) || duration, slot.start + 0.1), duration);
});
});
setState(); render(); return result;
})(secondsWidget.callback);
stateWidget.callback = ((previous) => function(...args) { const result = previous?.apply(this, args); render(); return result; })(stateWidget.callback);
const previousConnections = node.onConnectionsChange;
node.onConnectionsChange = function(...args) {
const result = previousConnections?.apply(this, args);
render();
return result;
};
node.__dumasSequencePromptBuilder = refresh;
refresh();
}
app.registerExtension({
name: "DUMAS.H3.SequencePromptBuilder",
nodeCreated(node) { setTimeout(() => configure(node), 0); },
loadedGraphNode(node) { setTimeout(() => configure(node), 0); },
afterConfigureGraph() { for (const node of app.graph?._nodes || []) configure(node); },
});