Add MiniMax H3 extender import and 3070 R2V variant

models/minimax/minimax-h3-extender-ref2va/README.md

@@ -0,0 +1,61 @@
+# MiniMax H3 Extender Ref2VA
+
+## Summary
+
+Imported `MiniMax H3` continuation workflow based on
+`tritant/ComfyUI_MiniMax_H3_Extender`. This is a `Ref2VA`-first extender with
+shared image references, optional audio reference, Motion Context continuity,
+validated clip-by-clip reuse, and final joined video/audio decode.
+
+## Model Family
+
+- `minimax`
+
+## Status
+
+- `imported-only`
+
+## Source
+
+- Repo: <https://github.com/tritant/ComfyUI_MiniMax_H3_Extender>
+- Workflow file: <https://github.com/tritant/ComfyUI_MiniMax_H3_Extender/blob/main/Workflow/MiniMax_Extender.json>
+- Original publisher: `tritant`
+- Date imported: `2026-08-16`
+
+## Developer notes
+
+- The upstream node is built around a single horizontal sequence node instead of
+  a large repeated graph.
+- It keeps a validated disk cache, supports `clip_by_clip` and `full_batch`,
+  and is explicitly designed for longer `MiniMax H3` continuation chains.
+- The upstream README says it combines `Ref2VA`, Motion Context, disk caching,
+  per-clip prompts, per-clip seeds, and final decode/export into a simpler
+  flow.
+
+## Our notes
+
+- This import is worth keeping because it is natively `reference-to-video`
+  oriented and closer to Chris's actual MiniMax use than generic `t2v` graphs.
+- I replaced the source repo's very specific demo clip script with a neutral
+  reusable starter clip state so the imported workflow opens as a practical
+  template instead of someone else's scene.
+- The graph still uses the upstream source-leaning stack, so it is more of a
+  donor / comparison import than the first `3070` workflow to run.
+
+## Required custom nodes
+
+- `ComfyUI_MiniMax_H3_Extender`
+
+## Required models
+
+- `MiniMax H3` `Ref2VA` checkpoint
+- matching `MiniMax H3` text encoder
+- `MiniMax H3` video VAE
+- `MiniMax H3` audio VAE
+- optional turbo LoRA path if enabled
+
+## Notes
+
+- Use this when you want the upstream extender logic itself.
+- For the tighter local-first variant, use the optimized companion workflow in
+  `optimized/image-to-video/minimax-h3-extender-3070-r2v-480p-clip-by-clip/`.

optimized/image-to-video/minimax-h3-extender-3070-r2v-480p-clip-by-clip/README.md

@@ -0,0 +1,61 @@
+# MiniMax H3 Extender 3070 R2V 480p Clip By Clip
+
+## Summary
+
+`MiniMax H3 Extender` optimized for Chris's actual MiniMax lane:
+`reference-to-video` first, incremental continuation, and safer `3070`
+execution. This variant keeps the extender's validated cache / clip approval
+model, but swaps it toward the lighter stack already used by the current local
+MiniMax extension masters.
+
+## Status
+
+- `best-current-extender-r2v-safe-start`
+
+## Workflow asset
+
+- `workflow.json`
+
+## Built from
+
+- `models/minimax/minimax-h3-extender-ref2va/`
+- upstream repo: `tritant/ComfyUI_MiniMax_H3_Extender`
+
+## Key model stack
+
+- `minimax_h3_ref2va_pruned_w4a8_mixed.safetensors`
+- `qwen3vl_32b_heretic_minimax_h3_nvfp4.safetensors`
+- `minimax_h3_video_vae_int8_convrot.safetensors`
+- `minimax_h3_audio_vae_fp32.safetensors`
+- `minimax\minimax_h3_fl2v_lightx2v_turbo_4step_v0.1_comfy.safetensors`
+
+## Main changes
+
+- switched the workflow to `clip_by_clip` so each clip can be generated,
+  reviewed, and validated before the chain continues
+- reduced the working canvas from `896x576` to `864x480`
+- raised the step count from `4` to `5` for a slightly less brittle low-step
+  path
+- swapped the source stack toward the lighter local stack already used by the
+  compact MiniMax continuation masters
+- replaced the source demo story with a neutral reusable `R2V` continuation
+  starter in `clips_json`
+
+## Why this exists
+
+- the upstream extender logic is genuinely useful, but the shipped example is
+  more optimistic than Chris's `3070 / 8 GB` box wants
+- Chris's primary MiniMax lane is `reference-to-video`, so the optimized
+  variant should open in an `R2V`-friendly continuation shape instead of
+  generic batch-generation mode
+- `clip_by_clip` plus validation is a better fit than `full_batch` when trying
+  to preserve continuity without wasting render time on bad downstream clips
+
+## Notes
+
+- This is the right workflow to test if the goal is "can the extender model be
+  useful on the local box?"
+- It does not replace the current compact `Contex Loop` extension masters yet;
+  it is the best direct `Extender`-style challenger to compare against them.
+- Start with one reference image, short `10s` clips, and explicit
+  `subject_definitions` blocks before scaling up.

optimized/notes/minimax.md

@@ -11,6 +11,7 @@
 - `models/minimax/minimax-h3-official-i2v/`
 - `models/minimax/minimax-h3-turbo-lora-community/`
 - `models/minimax/minimax-h3-int8-r2v-community/`
+- `models/minimax/minimax-h3-extender-ref2va/`
 
 ## Strong current heuristics
 
@@ -27,6 +28,7 @@
 
 - Keep `optimized/text-to-video/minimax-h3-3070-turbo-master/` as the first MiniMax default.
 - Keep the INT8 reference-video workflow as the next donor for future low-VRAM optimization.
+- Keep `optimized/image-to-video/minimax-h3-extender-3070-r2v-480p-clip-by-clip/` as the first `MiniMax H3 Extender` path to compare against the compact `Contex Loop` masters.
 
 ## Ref2VA prompt guidance