release: publish H3 Continuum V3.8X (3.8.1)

This commit is contained in:
ukr8b3g-cmyk
2026-09-15 03:46:57 +09:00
parent c421913bf1
commit 764179869f
62 changed files with 5707 additions and 2018 deletions
+395 -233
View File
@@ -6,9 +6,10 @@ runs for every chunk without intermediate VAE decoding, and decoding/assembly
happens after the sampling phase.
"""
from __future__ import annotations
import copy, hashlib, logging
from typing import Any
import torch
import copy, hashlib, logging
from typing import Any
import torch
from ..branch_provenance import physical_groups
from ..compatibility import accelerator_summary, check_comfy_h3_runtime
from ..constants import (
DIAGNOSTICS_FULL, DIAGNOSTICS_OFF, DIAGNOSTICS_OPTIONS, FPS,
@@ -30,7 +31,7 @@ from ..guide_timeline import (
validate_guide_visible_range,
)
from ..model_patch import LayoutValidationProfiler, clone_model_for_chunk
from ..state import assert_context_unchanged, context_fingerprint, make_plan, select_context, validate_state
from ..state import assert_context_unchanged, context_fingerprint, make_plan, select_context
from ..temporal import (
align_frame_count_up,
audio_latent_t,
@@ -40,6 +41,7 @@ from ..temporal import (
video_latent_t,
)
from ..version import PACKAGE_VERSION
from .decoder import decode_sequence, decode_sequence_with_seam, enforce_total_frames
from .context_diagnostics import ContextDiagnosticsTracker
from .h3_builder import attach_keyframes, empty_h3_latent, encode_identity_latents, encode_prompt_conditioning, encode_prompt_conditioning_cached, prepare_identity_assets
@@ -238,22 +240,6 @@ def _terminal_strategy_mismatch(
return saved_strategy != FLF_STRATEGY
def _terminal_sampling_plan(
*,
chunks: int,
completed: int,
merge_enabled: bool,
) -> tuple[tuple[int, ...], bool]:
if completed >= int(chunks):
return (), False
if not merge_enabled:
return tuple(range(completed, int(chunks))), False
pair_start = int(chunks) - 2
if completed > pair_start:
raise SequenceRuntimeError("terminal merged pair cannot resume from only one logical half")
return tuple(range(completed, pair_start)), True
def _physical_group_limit(value: int | None) -> int | None:
"""Validate the private V3 execution cap without changing sampling identity."""
@@ -264,9 +250,55 @@ def _physical_group_limit(value: int | None) -> int | None:
"max_new_physical_groups must be None or a positive integer"
)
return value
def _terminal_physical_seed_plan(
def _terminal_sampling_plan(
*,
chunks: int,
completed: int,
merge_enabled: bool,
) -> tuple[tuple[int, ...], bool]:
"""Legacy internal adapter delegated to the R3 Pure Planner."""
from ..v3.execution_planner import ExecutionPlanningError, build_execution_plan
from ..v3.planning_types import ContinuationSourceFacts
boundaries=tuple(
(int(group.start),int(group.end))
for group in physical_groups(
chunks=int(chunks),
terminal_merge_enabled=bool(merge_enabled),
)
if int(group.end)<=int(completed)
)
facts=ContinuationSourceFacts(
kind=("explicit_session" if int(completed)>0 else "none"),
accepted_chunks=int(completed),
reroll_from_chunk=0,
physical_group_boundaries=boundaries,
)
try:
plan=build_execution_plan(
continuation_source_facts=facts,
configured_chunks=int(chunks),
terminal_merge_enabled=bool(merge_enabled),
max_new_physical_groups=None,
review_execution=None,
)
except ExecutionPlanningError as exc:
raise SequenceRuntimeError(str(exc)) from exc
normal_indices=tuple(
descriptor.logical_chunks[0]-1
for descriptor in plan.groups_to_generate
if not descriptor.terminal_atomic
)
terminal_pending=any(
descriptor.terminal_atomic for descriptor in plan.groups_to_generate
)
return normal_indices,terminal_pending
def _terminal_physical_seed_plan(
*,
base_seed: int,
physical_window_start_index: int,
@@ -415,10 +447,22 @@ def _attach_terminal_flf_keyframes(
metadata["minimax_frame_count"] = int(frame_count)
return conditioning
def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,sigmas:torch.Tensor,first_frame:torch.Tensor|None,last_frame:torch.Tensor|None,prompt_plan:dict[str,Any],width:int,height:int,continuity:str,base_seed:int,audio_continuity:bool,exact_total_duration:bool,diagnostics_mode:str,reroll_from_chunk:int,reroll_nonce:int,strict_compatibility:bool,debug:bool,seam_correction:str=SEAM_CORRECTION_OFF,enable_preview:bool=True,session:dict[str,Any]|None=None,initial_state:dict[str,Any]|None=None,latent_only:bool=False,reference_assets=None,reference_audio_source=None,reference_audio_vae=None,driving_audio_source=None,driving_audio_vae=None,reference_video_source=None,timeline_video_source=None,guide_source=None,capture_refine_context:bool=False,memory_attribution:bool=False,prompt_conditioning_cache:bool=False,reference_encode_cache:bool=False,continuation_transport:str=REFERENCE_CONTEXT_V1,max_new_physical_groups:int|None=None,_memory_attribution_collector:Any=None,_diagnostic_continuation_policy:Any=None):
def _run_runtime_internal(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,sigmas:torch.Tensor,first_frame:torch.Tensor|None,last_frame:torch.Tensor|None,prompt_plan:dict[str,Any],width:int,height:int,continuity:str,base_seed:int,audio_continuity:bool,exact_total_duration:bool,diagnostics_mode:str,reroll_from_chunk:int,reroll_nonce:int,strict_compatibility:bool,debug:bool,seam_correction:str=SEAM_CORRECTION_OFF,enable_preview:bool=True,session:dict[str,Any]|None=None,initial_state:dict[str,Any]|None=None,latent_only:bool=False,reference_assets=None,reference_audio_source=None,reference_audio_vae=None,driving_audio_source=None,driving_audio_vae=None,reference_video_source=None,timeline_video_source=None,guide_source=None,capture_refine_context:bool=False,memory_attribution:bool=False,prompt_conditioning_cache:bool=False,reference_encode_cache:bool=False,continuation_transport:str=REFERENCE_CONTEXT_V1,max_new_physical_groups:int|None=None,_memory_attribution_collector:Any=None,_diagnostic_continuation_policy:Any=None):
from ..conditioning import detect_conditioning_mode, conditioning_display_label
from ..run_storage import get_active_run_storage
from ..v3.runtime_coordinator import (
InternalRuntimeCoordinator,
RuntimeCoordinatorError,
)
from ..v3.sampling_engine import sample_physical_group
storage_controller=get_active_run_storage()
runtime_coordinator=InternalRuntimeCoordinator(
storage_controller=storage_controller,
input_session=session,
initial_state=initial_state,
)
session,initial_state=runtime_coordinator.prepare_inputs()
review_execution=None
capture_refine_context=bool(capture_refine_context and latent_only)
memory_attribution=bool(memory_attribution and capture_refine_context)
prompt_conditioning_cache=bool(prompt_conditioning_cache and capture_refine_context and latent_only)
@@ -450,13 +494,13 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
# Legacy workflow input only. Runtime compatibility is advisory in V3.4.
strict_compatibility=False
if width<=0 or height<=0 or width%32 or height%32: raise SequenceRuntimeError("width and height must be positive multiples of 32")
if session is not None and initial_state is not None:
LOG.warning("Both session and initial_state were supplied; using the session and ignoring initial_state")
initial_state=None
if diagnostics_mode not in DIAGNOSTICS_OPTIONS: raise SequenceRuntimeError(f"unknown diagnostics mode: {diagnostics_mode!r}")
if seam_correction not in SEAM_CORRECTION_OPTIONS: raise SequenceRuntimeError(f"unknown seam correction mode: {seam_correction!r}")
if not 0<=int(reroll_from_chunk)<=chunks: raise SequenceRuntimeError("reroll_from_chunk must be 0 or a valid one-based chunk index")
if initial_state is not None and int(reroll_from_chunk) not in (0,1): raise SequenceRuntimeError("with initial_state, reroll_from_chunk can only be 0 or 1")
if diagnostics_mode not in DIAGNOSTICS_OPTIONS: raise SequenceRuntimeError(f"unknown diagnostics mode: {diagnostics_mode!r}")
if seam_correction not in SEAM_CORRECTION_OPTIONS: raise SequenceRuntimeError(f"unknown seam correction mode: {seam_correction!r}")
if not 0<=int(reroll_from_chunk)<=chunks: raise SequenceRuntimeError("reroll_from_chunk must be 0 or a valid one-based chunk index")
try:
runtime_coordinator.validate_initial_state_reroll(reroll_from_chunk)
except RuntimeCoordinatorError as exc:
raise SequenceRuntimeError(str(exc)) from exc
try: conditioning_mode=detect_conditioning_mode(first_frame=first_frame,last_frame=last_frame,reference_assets=reference_assets)
except ValueError as exc: raise SequenceRuntimeError(str(exc)) from exc
conditioning_display=conditioning_display_label(has_first=first_frame is not None,has_last=last_frame is not None,has_reference=reference_assets is not None)
@@ -500,8 +544,9 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
if multi_chunk_flf:
sequence_identity_hash=_flf_contract_identity(sequence_identity_hash)
sequence_identity_hash=_continuation_transport_identity(sequence_identity_hash,continuation_transport)
current_model_fingerprint=model_fingerprint(model,extra_wrapper_keys=("h3_continuum_join.apply_model.v1",))
if storage_controller is not None:
current_model_fingerprint=model_fingerprint(model,extra_wrapper_keys=("h3_continuum_join.apply_model.v1",))
if storage_controller is not None:
runtime_coordinator.assert_storage_session_compatible()
stored_session=storage_controller.prepare(model=model,model_fingerprint_value=current_model_fingerprint,clip=clip,video_vae=video_vae,sampler=sampler,sigmas=sigmas,prompt_plan=plan,width=width,height=height,chunk_seconds=chunk_seconds,continuity=continuity,audio_continuity=audio_continuity,base_seed=base_seed,reroll_from_chunk=reroll_from_chunk,reroll_nonce=reroll_nonce,first_frame_hash=assets.first_frame_hash,last_frame_hash=assets.last_frame_hash,identity_hash=sequence_identity_hash,strict_compatibility=strict_compatibility,existing_session=session,reference_contract=reference_assets.contract if reference_assets is not None else None,conditioning_mode=conditioning_mode,reference_audio_contract=reference_audio_source.contract if reference_audio_source is not None else None,reference_audio_vae=reference_audio_vae,driving_audio_contract=driving_audio_source.contract if driving_audio_source is not None else None,driving_audio_vae=driving_audio_vae,reference_video_contract=reference_video_source.contract if reference_video_source is not None else None,timeline_video_contract=timeline_video_source.contract if timeline_video_source is not None else None,guide_contract=guide_source.contract if guide_source is not None else None,execution_semantics=_terminal_execution_semantics(merge_enabled=terminal_merge_enabled,prompt_policy=terminal_prompt_policy,continuation_transport=continuation_transport))
reroll_nonce=storage_controller.effective_reroll_nonce
review_execution=getattr(storage_controller,"review_execution",None)
@@ -579,12 +624,62 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
f"{guide_source.absolute_frame}; mode={guide_source.contract['mode']}; "
f"schema={guide_source.contract['schema_version']}.",
)
if session is not None:
old_fingerprint=str(session.get("model_fingerprint",""))
if old_fingerprint and old_fingerprint!=current_model_fingerprint: reuse_notes.append("model/accelerator fingerprint differs from the saved session; accepted chunks were kept")
if session is not None:
old_fingerprint=str(session.get("model_fingerprint",""))
if old_fingerprint and old_fingerprint!=current_model_fingerprint: reuse_notes.append("model/accelerator fingerprint differs from the saved session; accepted chunks were kept")
# R4: the Coordinator is the sole owner of Storage / Session / State
# priority. The Pure Planner receives only its selected-source projection.
try:
continuation_selection=runtime_coordinator.select_continuation_source(
preserved=preserved,
initial_state=initial_state,
chunks=chunks,
width=width,
height=height,
reroll_from_chunk=int(reroll_from_chunk),
effective_reroll_from_chunk=int(effective_reroll_from_chunk),
terminal_merge_enabled=terminal_merge_enabled,
)
except RuntimeCoordinatorError as exc:
raise SequenceRuntimeError(str(exc)) from exc
entries=continuation_selection.entries
previous_state=continuation_selection.previous_state
continuation_source_facts=continuation_selection.continuation_source_facts
reuse_notes.extend(continuation_selection.notes)
retained_review_unit=None
inherited_review_unit=(
getattr(storage_controller,"inherited_review_unit",None)
if storage_controller is not None
else None
)
if isinstance(inherited_review_unit,dict):
from ..v3.review_control import ReviewUnit
retained_review_unit=ReviewUnit(
start=int(inherited_review_unit["start"]),
end=int(inherited_review_unit["end"]),
physical_group=int(inherited_review_unit["physical_group"]),
)
from ..v3.execution_planner import ExecutionPlanningError,build_execution_plan
try:
execution_plan=build_execution_plan(
continuation_source_facts=continuation_source_facts,
configured_chunks=chunks,
terminal_merge_enabled=terminal_merge_enabled,
max_new_physical_groups=max_new_physical_groups,
review_execution=review_execution,
retained_review_unit=retained_review_unit,
)
except ExecutionPlanningError as exc:
raise SequenceRuntimeError(str(exc)) from exc
if storage_controller is not None:
# Queue-local planning result only. No Run Storage schema or persisted
# metadata is changed in R3.
storage_controller.execution_plan=execution_plan
cache={}
guide_assets=None
if len(preserved)<chunks:
if len(preserved)<chunks or capture_refine_context:
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="conditioning")
capture_memory(memory_collector, "start_phase", phase="conditioning_identity_video_vae")
@@ -651,8 +746,37 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
if timeline_video_source is None:
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="conditioning_prompt_clip")
cache=_conditioning_cache(clip=clip,prompts=prompts,assets=assets,final_has_last_frame=last_frame is not None,reference_assets=reference_assets,reference_audio_assets=reference_audio_assets,timeline_video_assets=reference_video_assets,prompt_conditioning_cache=prompt_conditioning_cache,prompt_cache_event=prompt_cache_events.append)
if terminal_merge_enabled and terminal_prompt is not None:
planned_normal_indices=tuple(
descriptor.logical_chunks[0]-1
for descriptor in execution_plan.groups_to_generate
if not descriptor.terminal_atomic
)
if max_new_physical_groups is None:
# Preserve the established Full Run conditioning/cache set,
# including both logical prompts covered by Terminal Merge.
prepared_indices=tuple(range(chunks))
elif capture_refine_context:
normal_stop=chunks-2 if terminal_merge_enabled else chunks
reused_normal_indices=tuple(range(min(len(entries),normal_stop)))
prepared_indices=tuple(dict.fromkeys(
reused_normal_indices+planned_normal_indices
))
else:
prepared_indices=planned_normal_indices
prepared_prompts=[prompts[index] for index in prepared_indices]
prepare_last=bool(last_frame is not None and (
max_new_physical_groups is None or chunks-1 in prepared_indices
))
prepare_terminal=any(
descriptor.terminal_atomic
for descriptor in execution_plan.groups_to_generate
) or bool(
capture_refine_context
and terminal_merge_enabled
and len(entries)==chunks
)
cache=_conditioning_cache(clip=clip,prompts=prepared_prompts,assets=assets,final_has_last_frame=prepare_last,reference_assets=reference_assets,reference_audio_assets=reference_audio_assets,timeline_video_assets=reference_video_assets,prompt_conditioning_cache=prompt_conditioning_cache,prompt_cache_event=prompt_cache_events.append)
if prepare_terminal and terminal_prompt is not None:
terminal_key=_conditioning_cache_key(terminal_prompt,include_last=True,reference_assets=reference_assets)
if terminal_key not in cache:
cache.update(_conditioning_cache(clip=clip,prompts=[terminal_prompt],assets=assets,final_has_last_frame=True,reference_assets=reference_assets,reference_audio_assets=reference_audio_assets,timeline_video_assets=reference_video_assets,prompt_conditioning_cache=prompt_conditioning_cache,prompt_cache_event=prompt_cache_events.append))
@@ -660,21 +784,7 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
capture_memory(memory_collector, "finish_phase", phase="conditioning_prompt_clip")
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="conditioning")
entries=preserved[:]; previous_state=None
if entries:
try: previous_state=entry_to_state(entries[-1])
except (SessionValidationError, ValueError) as exc:
reuse_notes.append(f"saved continuation state was rejected; generated a fresh run ({exc})")
entries=[]; previous_state=None
elif initial_state is not None:
try:
candidate=validate_state(initial_state)
if int(candidate["width"])!=width or int(candidate["height"])!=height:
reuse_notes.append("initial_state resolution differs; generated a fresh run")
else: previous_state=candidate
except ValueError as exc:
reuse_notes.append(f"initial_state was rejected; generated a fresh run ({exc})")
initial_frame_count=align_frame_count_up(int(round(chunk_seconds*FPS))); retained_frames=sum(int(entry["plan"]["net_frames"]) for entry in entries); sampling_reports=[]
initial_frame_count=align_frame_count_up(int(round(chunk_seconds*FPS))); retained_frames=sum(int(entry["plan"]["net_frames"]) for entry in entries); sampling_reports=[]
context_diagnostics=ContextDiagnosticsTracker() if bool(debug) else None
if context_diagnostics is not None:
if entries:
@@ -682,10 +792,26 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
_record_context_diagnostics(tracker=context_diagnostics,reports=sampling_reports,state=entry_to_state(reused_entry),continuity=continuity,reused=True)
elif previous_state is not None:
_record_context_diagnostics(tracker=context_diagnostics,reports=sampling_reports,state=previous_state,continuity=continuity,reused=True)
normal_indices,terminal_merge_pending=_terminal_sampling_plan(chunks=chunks,completed=len(entries),merge_enabled=terminal_merge_enabled)
if max_new_physical_groups == 0:
normal_indices=()
terminal_merge_pending=False
normal_indices=tuple(
descriptor.logical_chunks[0]-1
for descriptor in execution_plan.groups_to_generate
if not descriptor.terminal_atomic
)
terminal_merge_pending=any(
descriptor.terminal_atomic
for descriptor in execution_plan.groups_to_generate
)
# Reuse the same group preparation for context capture; never resample or
# recommit accepted entries. An external initial State is not reconstructible
# from the saved prefix alone, so retain incomplete-context behavior for it.
reused_count=len(entries)
rebuild_prefix=bool(capture_refine_context and entries and not entries[0]["plan"]["continuation"])
terminal_reused=bool(rebuild_prefix and terminal_merge_enabled and reused_count==chunks)
if rebuild_prefix:
normal_stop=chunks-2 if terminal_merge_enabled else chunks
normal_indices=tuple(range(min(reused_count,normal_stop)))+tuple(normal_indices)
previous_state=None
retained_frames=0
if adaptive_observer_event is not None:
adaptive_observer_event(
adaptive_continuity_observer,
@@ -699,6 +825,7 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="preparation")
for sequence_index in normal_indices:
reusing_group=sequence_index<reused_count
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="group_prepare", physical_group=sequence_index+1, logical_chunks=(sequence_index+1,))
prompt=prompts[sequence_index]; prompt_hash_value=prompt_hashes[sequence_index]; is_final=sequence_index==chunks-1; effective_reroll_nonce=int(reroll_nonce) if int(reroll_from_chunk)>0 and sequence_index+1>=int(reroll_from_chunk) else 0; seed=derive_chunk_seed(base_seed,sequence_index,effective_reroll_nonce); motion_score=0.0; video_context=None; audio_context=None; context_before=None; context_interop_emitted=False
@@ -776,7 +903,7 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
conditioning=prepare_conditioning(base_conditioning,video_context=video_context,audio_context=audio_context,audio_grid_offset=grid_offset,context_frames=context_frames,new_frame_count=shape.total_frames,first_frame_policy=POLICY_REPLACE,preserve_last_frame=True)
context_interop_emitted=True
clip_index=int(previous_state["clip_index"])+1; chunk_plan=make_plan(continuation=True,clip_index=clip_index,total_frames=shape.total_frames,trim_frames=context_frames,width=width,height=height,context_frames=context_frames,state_capacity_frames=largest_context_capacity(shape.net_new_frames),requested_extend_seconds=chunk_seconds,debug=debug)
if adaptive_observer_event is not None:
if adaptive_observer_event is not None and not reusing_group:
from ..v3.adaptive_continuity import INITIAL_TRANSPORT
adaptive_observer_event(
adaptive_continuity_observer,
@@ -821,7 +948,8 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
clone_kwargs={"strict":bool(strict_compatibility),"debug":bool(debug),"chunk_index":clip_index,"context_frames":context_frames if context_interop_emitted else None}
if layout_validation_profiler is not None: clone_kwargs["validation_profiler"]=layout_validation_profiler
if packed_row_planner is not None: clone_kwargs["packed_row_planner"]=packed_row_planner
chunk_model=clone_model_for_chunk(model,**clone_kwargs)
if not reusing_group:
chunk_model=clone_model_for_chunk(model,**clone_kwargs)
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="group_prepare", physical_group=sequence_index+1, logical_chunks=(sequence_index+1,))
capture_memory(memory_collector, "start_phase", phase="refine_context", physical_group=sequence_index+1, logical_chunks=(sequence_index+1,))
@@ -855,6 +983,13 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
del source_video,_
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="refine_context", physical_group=sequence_index+1, logical_chunks=(sequence_index+1,))
if reusing_group:
previous_state=entry_to_state(entries[sequence_index])
retained_frames+=int(entries[sequence_index]["plan"]["net_frames"])
del latent,conditioning,chunk_cache
if timeline_video_source is not None:
del timeline_video_assets
continue
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=sequence_index+1,
@@ -866,39 +1001,24 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
)
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="sampling", physical_group=sequence_index+1, logical_chunks=(sequence_index+1,))
layout_profile_token=(
layout_validation_profiler.begin_sampling_group(
physical_group=sequence_index+1,
logical_chunks=(sequence_index+1,),
sampling_steps=max(0,int(sigmas.shape[-1])-1),
)
if layout_validation_profiler is not None
else None
sampled,diagnostic_sample_token=sample_physical_group(
sample_callable=sample_chunk,
model=chunk_model,
conditioning=conditioning,
latent=latent,
sampler=sampler,
sigmas=sigmas,
seed=seed,
enable_preview=bool(enable_preview),
physical_group=sequence_index+1,
logical_chunks=(sequence_index+1,),
terminal_atomic=False,
layout_validation_profiler=layout_validation_profiler,
packed_row_planner=packed_row_planner,
planner_event=planner_event,
diagnostic_policy=_diagnostic_continuation_policy,
context_frames=context_frames,
)
diagnostic_sample_token=None
if _diagnostic_continuation_policy is not None:
diagnostic_sample_token=_diagnostic_continuation_policy.before_sampling(
physical_group=sequence_index+1,
logical_chunks=(sequence_index+1,),
context_frames=context_frames,
latent=latent,
conditioning=conditioning,
seed=seed,
sigmas=sigmas,
)
if planner_event is not None:
planner_event(
packed_row_planner,
"begin_group",
physical_group=sequence_index+1,
logical_chunks=(sequence_index+1,),
terminal_atomic=False,
)
sampled=sample_chunk(model=chunk_model,conditioning=conditioning,latent=latent,sampler=sampler,sigmas=sigmas,seed=seed,enable_preview=bool(enable_preview))
if planner_event is not None:
planner_event(packed_row_planner,"finish_group")
if layout_validation_profiler is not None and layout_profile_token is not None:
layout_validation_profiler.finish_sampling_group(layout_profile_token)
if previous_state is not None and continuation_transport in (MASKED_VIDEO_PREFIX_V1,MASKED_AV_PREFIX_22_V1,MASKED_AV_PREFIX_39_V1):
if continuation_transport in (MASKED_AV_PREFIX_22_V1,MASKED_AV_PREFIX_39_V1):
from ..v3.masked_continuation import restore_masked_av_prefix
@@ -933,7 +1053,7 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
if context_before is not None and video_context is not None: assert_context_unchanged(video_context,audio_context,context_before)
entry=make_chunk_entry(latent=sampled,plan=chunk_plan,prompt=prompt,prompt_hash=prompt_hash_value,seed=seed,context_frames=context_frames,motion_score=motion_score,reused=False); previous_state=entry_to_state(entry); entries.append(entry)
_record_context_diagnostics(tracker=context_diagnostics,reports=sampling_reports,state=previous_state,continuity=continuity,reused=False)
if storage_controller is not None: storage_controller.commit_chunk(entry, position=sequence_index)
if storage_controller is not None: storage_controller.commit_group((entry,), positions=(sequence_index,))
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="cpu_commit_validation", physical_group=sequence_index+1, logical_chunks=(sequence_index+1,))
retained_frames+=int(chunk_plan["net_frames"])
@@ -956,18 +1076,9 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
physical_group=sequence_index+1,
)
new_physical_groups+=1
if (
max_new_physical_groups is not None
and new_physical_groups>=max_new_physical_groups
):
break
terminal_budget_available=(
max_new_physical_groups is None
or new_physical_groups<max_new_physical_groups
)
if terminal_merge_pending and terminal_budget_available:
if terminal_reused or terminal_merge_pending:
pair_start=chunks-2
if len(entries)!=pair_start:
if len(entries)!=(chunks if terminal_reused else pair_start):
raise SequenceRuntimeError(f"terminal merged pair expected {pair_start} completed chunks, got {len(entries)}")
terminal_physical_group=pair_start+1
terminal_logical_chunks=(pair_start+1,chunks)
@@ -1044,7 +1155,7 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
conditioning=prepare_conditioning(base_conditioning,video_context=video_context,audio_context=audio_context,audio_grid_offset=grid_offset,context_frames=TERMINAL_MERGE_CONTEXT_FRAMES,new_frame_count=physical_frames,first_frame_policy=POLICY_REPLACE,preserve_last_frame=True)
reason=f"terminal merged 10-second sample with 22-frame context; continuity selection was {selected_context} ({selected_reason})"
physical_clip_index=int(previous_state["clip_index"])+1
if adaptive_observer_event is not None:
if adaptive_observer_event is not None and not terminal_reused:
from ..v3.adaptive_continuity import INITIAL_TRANSPORT
adaptive_observer_event(
adaptive_continuity_observer,
@@ -1090,27 +1201,28 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
)
driving_audio_latent=slice_driving_audio_latent(driving_audio_assets,cumulative_retained_before=retained_frames,total_frames=physical_frames,trim_frames=physical_context_frames,fps=FPS)
conditioning=attach_driving_audio(conditioning,driving_audio_latent)
chunk_model=clone_model_for_chunk(
model,
strict=bool(strict_compatibility),
debug=bool(debug),
chunk_index=physical_clip_index,
context_frames=(
physical_context_frames
if not initial_pair and not terminal_masked
else None
),
**(
{"validation_profiler": layout_validation_profiler}
if layout_validation_profiler is not None
else {}
),
**(
{"packed_row_planner": packed_row_planner}
if packed_row_planner is not None
else {}
),
)
if not terminal_reused:
chunk_model=clone_model_for_chunk(
model,
strict=bool(strict_compatibility),
debug=bool(debug),
chunk_index=physical_clip_index,
context_frames=(
physical_context_frames
if not initial_pair and not terminal_masked
else None
),
**(
{"validation_profiler": layout_validation_profiler}
if layout_validation_profiler is not None
else {}
),
**(
{"packed_row_planner": packed_row_planner}
if packed_row_planner is not None
else {}
),
)
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="group_prepare", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
capture_memory(memory_collector, "start_phase", phase="refine_context", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
@@ -1144,111 +1256,115 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
del source_video,_
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="refine_context", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
stage="before sampling",
retained_entries=entries,
observed_latent=latent,
retained_refine_context=refine_groups,
)
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="sampling", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
layout_profile_token=(
layout_validation_profiler.begin_sampling_group(
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
sampling_steps=max(0,int(sigmas.shape[-1])-1),
)
if layout_validation_profiler is not None
else None
)
if planner_event is not None:
planner_event(
packed_row_planner,
"begin_group",
if not terminal_reused:
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
stage="before sampling",
retained_entries=entries,
observed_latent=latent,
retained_refine_context=refine_groups,
)
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="sampling", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
sampled,_=sample_physical_group(
sample_callable=sample_chunk,
model=chunk_model,
conditioning=conditioning,
latent=latent,
sampler=sampler,
sigmas=sigmas,
seed=physical_seed,
enable_preview=bool(enable_preview),
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
terminal_atomic=True,
layout_validation_profiler=layout_validation_profiler,
packed_row_planner=packed_row_planner,
planner_event=planner_event,
)
sampled=sample_chunk(model=chunk_model,conditioning=conditioning,latent=latent,sampler=sampler,sigmas=sigmas,seed=physical_seed,enable_preview=bool(enable_preview))
if planner_event is not None:
planner_event(packed_row_planner,"finish_group")
if layout_validation_profiler is not None and layout_profile_token is not None:
layout_validation_profiler.finish_sampling_group(layout_profile_token)
if terminal_masked:
if continuation_transport==MASKED_AV_PREFIX_22_V1:
from ..v3.masked_continuation import restore_masked_av_prefix
sampled=restore_masked_av_prefix(
sampled,
latent,
TERMINAL_MERGE_CONTEXT_FRAMES,
if terminal_masked:
if continuation_transport==MASKED_AV_PREFIX_22_V1:
from ..v3.masked_continuation import restore_masked_av_prefix
sampled=restore_masked_av_prefix(
sampled,
latent,
TERMINAL_MERGE_CONTEXT_FRAMES,
)
else:
from ..v3.masked_continuation import restore_masked_video_prefix
sampled=restore_masked_video_prefix(
sampled,
latent,
TERMINAL_MERGE_CONTEXT_FRAMES,
)
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="sampling", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
stage="after sampling",
retained_entries=entries,
observed_latent=sampled,
retained_refine_context=refine_groups,
)
else:
from ..v3.masked_continuation import restore_masked_video_prefix
sampled=restore_masked_video_prefix(
sampled,
latent,
TERMINAL_MERGE_CONTEXT_FRAMES,
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="cpu_commit_validation", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
if context_before is not None and video_context is not None:
assert_context_unchanged(video_context,audio_context,context_before)
sampled_video,sampled_audio=latent_to_cpu(sampled)
logical_parts=_split_terminal_merged_latents(sampled_video,sampled_audio,contract)
sampling_reports.append(f"terminal physical sample: terminal_pair={pair_start+1}-{chunks}, physical_seed={physical_seed}, physical_frames={physical_frames}, sampling_passes=1, shared_physical_sample=true, terminal_prompt_policy={terminal_prompt_policy}, trim={physical_context_frames}, reason={reason}")
terminal_commit_entries=[]
terminal_commit_positions=[]
for offset,(video_part,audio_part) in enumerate(logical_parts):
sequence_index=pair_start+offset
total_frames=int(contract["logical_frames"][offset])
trim_frames=int(contract["logical_trims"][offset])
continuation=bool(trim_frames)
clip_index=physical_clip_index+offset
context_frames=trim_frames if continuation else 5
chunk_plan=make_plan(continuation=continuation,clip_index=clip_index,total_frames=total_frames,trim_frames=trim_frames,width=width,height=height,context_frames=context_frames,state_capacity_frames=largest_context_capacity(total_frames-trim_frames),requested_extend_seconds=chunk_seconds,debug=debug)
logical_latent=latent_from_cpu(video_part,audio_part)
logical_seed=int(terminal_seed_plan["logical_entry_seeds"][offset])
entry=make_chunk_entry(latent=logical_latent,plan=chunk_plan,prompt=prompts[sequence_index],prompt_hash=prompt_hashes[sequence_index],seed=logical_seed,context_frames=trim_frames,motion_score=motion_score,reused=False)
entries.append(entry)
previous_state=entry_to_state(entry)
_record_context_diagnostics(tracker=context_diagnostics,reports=sampling_reports,state=previous_state,continuity=continuity,reused=False)
terminal_commit_entries.append(entry)
terminal_commit_positions.append(sequence_index)
retained_frames+=int(chunk_plan["net_frames"])
sampling_reports.append(f"chunk {sequence_index+1}/{chunks}: seed={logical_seed}, frames={total_frames}, trim={trim_frames}, retained_total={retained_frames}, context={trim_frames}, shared_physical_sample=terminal_10s_seed_v2")
del logical_latent
if storage_controller is not None:
storage_controller.commit_group(
tuple(terminal_commit_entries),
positions=tuple(terminal_commit_positions),
)
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="sampling", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
stage="after sampling",
retained_entries=entries,
observed_latent=sampled,
retained_refine_context=refine_groups,
)
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="cpu_commit_validation", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
if context_before is not None and video_context is not None:
assert_context_unchanged(video_context,audio_context,context_before)
sampled_video,sampled_audio=latent_to_cpu(sampled)
logical_parts=_split_terminal_merged_latents(sampled_video,sampled_audio,contract)
sampling_reports.append(f"terminal physical sample: terminal_pair={pair_start+1}-{chunks}, physical_seed={physical_seed}, physical_frames={physical_frames}, sampling_passes=1, shared_physical_sample=true, terminal_prompt_policy={terminal_prompt_policy}, trim={physical_context_frames}, reason={reason}")
for offset,(video_part,audio_part) in enumerate(logical_parts):
sequence_index=pair_start+offset
total_frames=int(contract["logical_frames"][offset])
trim_frames=int(contract["logical_trims"][offset])
continuation=bool(trim_frames)
clip_index=physical_clip_index+offset
context_frames=trim_frames if continuation else 5
chunk_plan=make_plan(continuation=continuation,clip_index=clip_index,total_frames=total_frames,trim_frames=trim_frames,width=width,height=height,context_frames=context_frames,state_capacity_frames=largest_context_capacity(total_frames-trim_frames),requested_extend_seconds=chunk_seconds,debug=debug)
logical_latent=latent_from_cpu(video_part,audio_part)
logical_seed=int(terminal_seed_plan["logical_entry_seeds"][offset])
entry=make_chunk_entry(latent=logical_latent,plan=chunk_plan,prompt=prompts[sequence_index],prompt_hash=prompt_hashes[sequence_index],seed=logical_seed,context_frames=trim_frames,motion_score=motion_score,reused=False)
entries.append(entry)
previous_state=entry_to_state(entry)
_record_context_diagnostics(tracker=context_diagnostics,reports=sampling_reports,state=previous_state,continuity=continuity,reused=False)
if storage_controller is not None:
storage_controller.commit_chunk(entry,position=sequence_index)
retained_frames+=int(chunk_plan["net_frames"])
sampling_reports.append(f"chunk {sequence_index+1}/{chunks}: seed={logical_seed}, frames={total_frames}, trim={trim_frames}, retained_total={retained_frames}, context={trim_frames}, shared_physical_sample=terminal_10s_seed_v2")
del logical_latent
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="cpu_commit_validation", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
del sampled,sampled_video,sampled_audio,logical_parts,latent,conditioning,chunk_model
del video_part,audio_part
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
stage="after CPU commit",
retained_entries=entries,
retained_refine_context=refine_groups,
)
mark_review_group=getattr(storage_controller,"mark_review_group",None)
if callable(mark_review_group):
mark_review_group(
start=pair_start+1,
end=chunks,
physical_group=terminal_physical_group,
)
new_physical_groups+=1
if memory_collector is not None:
capture_memory(memory_collector, "finish_phase", phase="cpu_commit_validation", physical_group=terminal_physical_group, logical_chunks=terminal_logical_chunks)
del sampled,sampled_video,sampled_audio,logical_parts,latent,conditioning,chunk_model
del video_part,audio_part
if memory_collector is not None:
capture_memory(memory_collector, "capture_group",
physical_group=terminal_physical_group,
logical_chunks=terminal_logical_chunks,
stage="after CPU commit",
retained_entries=entries,
retained_refine_context=refine_groups,
)
mark_review_group=getattr(storage_controller,"mark_review_group",None)
if callable(mark_review_group):
mark_review_group(
start=pair_start+1,
end=chunks,
physical_group=terminal_physical_group,
)
new_physical_groups+=1
else:
del latent,conditioning
if memory_collector is not None:
capture_memory(memory_collector, "start_phase", phase="finalization")
if layout_validation_profiler is not None:
@@ -1351,11 +1467,57 @@ def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,si
if duration_report: report_lines.append(duration_report)
report_lines.extend([session_summary(new_session),f"Output: {images.shape[0]} frames ({images.shape[0]/FPS:.3f}s), audio samples={audio['waveform'].shape[-1]}, decoded tensor≈{decoded_gib:.2f} GiB."])
return images,audio,last_state,new_session,"\n".join(report_lines)
# V3.0.1 hardening integration: Detailed Report only; generation semantics unchanged.
from ..hardening import run_sequence_with_hardening as _run_sequence_with_hardening
_run_sequence_v300 = run_sequence
def run_sequence(*args, **kwargs):
return _run_sequence_with_hardening(_run_sequence_v300, args, kwargs)
# V3.0.1 hardening integration: Detailed Report only; generation semantics unchanged.
from ..hardening import run_sequence_with_hardening as _run_sequence_with_hardening
def run_sequence(*,model:Any,clip:Any,video_vae:Any,audio_vae:Any,sampler:Any,sigmas:torch.Tensor,first_frame:torch.Tensor|None,last_frame:torch.Tensor|None,prompt_plan:dict[str,Any],width:int,height:int,continuity:str,base_seed:int,audio_continuity:bool,exact_total_duration:bool,diagnostics_mode:str,reroll_from_chunk:int,reroll_nonce:int,strict_compatibility:bool,debug:bool,seam_correction:str=SEAM_CORRECTION_OFF,enable_preview:bool=True,session:dict[str,Any]|None=None,initial_state:dict[str,Any]|None=None,latent_only:bool=False,reference_assets=None,reference_audio_source=None,reference_audio_vae=None,driving_audio_source=None,driving_audio_vae=None,reference_video_source=None,timeline_video_source=None,guide_source=None,capture_refine_context:bool=False,memory_attribution:bool=False,prompt_conditioning_cache:bool=False,reference_encode_cache:bool=False,continuation_transport:str=REFERENCE_CONTEXT_V1,max_new_physical_groups:int|None=None,_memory_attribution_collector:Any=None,_diagnostic_continuation_policy:Any=None):
"""Signature-compatible adapter into the R4 internal runtime."""
return _run_sequence_with_hardening(
_run_runtime_internal,
(),
{
"model":model,
"clip":clip,
"video_vae":video_vae,
"audio_vae":audio_vae,
"sampler":sampler,
"sigmas":sigmas,
"first_frame":first_frame,
"last_frame":last_frame,
"prompt_plan":prompt_plan,
"width":width,
"height":height,
"continuity":continuity,
"base_seed":base_seed,
"audio_continuity":audio_continuity,
"exact_total_duration":exact_total_duration,
"diagnostics_mode":diagnostics_mode,
"reroll_from_chunk":reroll_from_chunk,
"reroll_nonce":reroll_nonce,
"strict_compatibility":strict_compatibility,
"debug":debug,
"seam_correction":seam_correction,
"enable_preview":enable_preview,
"session":session,
"initial_state":initial_state,
"latent_only":latent_only,
"reference_assets":reference_assets,
"reference_audio_source":reference_audio_source,
"reference_audio_vae":reference_audio_vae,
"driving_audio_source":driving_audio_source,
"driving_audio_vae":driving_audio_vae,
"reference_video_source":reference_video_source,
"timeline_video_source":timeline_video_source,
"guide_source":guide_source,
"capture_refine_context":capture_refine_context,
"memory_attribution":memory_attribution,
"prompt_conditioning_cache":prompt_conditioning_cache,
"reference_encode_cache":reference_encode_cache,
"continuation_transport":continuation_transport,
"max_new_physical_groups":max_new_physical_groups,
"_memory_attribution_collector":_memory_attribution_collector,
"_diagnostic_continuation_policy":_diagnostic_continuation_policy,
},
)