The 3-step distill (speedup) LoRA is now baked into the diffusion model
itself, so the Moxie Multimedia Loader no longer applies any LoRA:
- moxie_loader.py: drop the models/loras scan, _auto_lora(), the module
level LoRA cache and _apply_lora(); the load chain is now
diffusion model -> attention backend -> CLIP -> VAEs. A leftover file
in models/loras is silently ignored and the folder can be deleted.
- moxie_project.py / moxie_preview.py: tooltips reference the baked-in
distill instead of the LoRA.
- tests + standalone loader script: assert no load_lora_for_models call
happens (the comfy.sd stub now raises if the loader tries), and fix
the pre-existing MMnode.tiny_vae stub gap (missing pack_vae_approx_files)
that broke the whole test module at fixture setup.
- README: models table drops the Turbo LoRA row; CHANGELOG entry added.
Node schema, outputs and sampling behavior are unchanged; workflows
saved before this change keep working.
PyPI only hosts a stub sdist for nvidia-vfx whose wheel_stub build
backend aborts the entire 'pip install -r requirements.txt' run
(BackendUnavailable: Cannot import 'wheel_stub.buildapi'). The real
prebuilt wheels (win_amd64 cp310/cp311/cp312-abi3 + manylinux) live on
pypi.nvidia.com. Route nvidia-vfx there with file-level pip options
(--extra-index-url + --only-binary=nvidia-vfx) so pip always picks the
wheel and the stub is never built. Validated in a clean venv: the
wheel installs and provides the nvvfx module rtx_vsr.py imports.
HF shows 'empty or missing yaml metadata in repo card' when the
README has no YAML front matter. Add license (gpl-3.0) and discovery
tags at the top of README.md, matching the structure Comfy-Org uses
on its repackaged model repos.
The pack targets NVIDIA RTX 3060+ GPUs only, so speech (voxcpm),
subtitle recognition (openai-whisper, qwen-asr) and the RTX VSR
upscale (nvidia-vfx) move from on-demand extras into requirements.txt.
Verified on PyPI: all four resolve, torchcodec ships Windows wheels
for python 3.10-3.14, and the qwen-asr transformers==4.57.6 pin is
compatible with the rest of the set. requirements.txt is ordered
light -> heavy so re-running the install resumes cleanly.
The downloader script is removed; all model files live in the Hugging
Face version of the repository, so cloning it is the only step needed.
CHANGELOG also records the new requirements.txt from the previous
commit.
ComfyUI-Manager installs requirements.txt automatically, but the pack
never shipped one, so manual/portable installs got zero dependencies.
List the feature-level deps (soundfile, jieba, huggingface_hub,
ffmpeg-python, opencv-python, imageio) and document the embedded-python
install command for ComfyUI Portable plus optional extras (voxcpm,
openai-whisper, qwen-asr, nvidia-vfx).
- T2V selections (editor default mode, explicit [start-end,t2v] override
tags, raw t2v values) now resolve to r2v in every task-type resolver
(h3_task_type, _multitrack_task_type, frontend getMultiTrackTaskType);
h3_generation_mode maps raw t2v to reference conditioning as belt and
braces. Reference-mode with zero reference media renders text-only
locally; the cloud path falls back to its native t2va request instead
of raising. i2v/l2v/v2v/edit flows unchanged.
- MoxieMultimediaLoader now presents a single model picker: LoRA, video
VAE, audio VAE, and CLIP widgets removed and auto-selected from the
pack's models/ folders (first sorted file, loud error when missing,
LoRA applied at strength 1.0 when bundled). Outputs and load order
unchanged; saved workflows keep working (extra widget values ignored).
Verified: py_compile; scripts/test_t2v_r2v_and_loader.py 50/50;
frontend vitest 566/567 (one pre-existing env failure on HEAD);
bun run build; scripts/test_rtx_upscale_integration.py all pass.
MoxieCombine now exposes exactly images / audio / fps - the leftover
VIDEO and FILENAME_PREFIX outputs are gone, so the output file name is
chosen on easy saveVideo itself. The base easy multitrackProjectVideoCombine
keeps its VIDEO and FILENAME_PREFIX outputs for legacy workflows.
The shipped Moxie-consolidated template is rewired accordingly: fps/images
links follow the new slot indices (2/0), the filename_prefix link is replaced
by the same project-scoped literal the removed output produced, and the
TEMPLATE's save location is unchanged.
pytest: new schema/execute/blocker coverage for MoxieCombine via a
_load_moxie_combine helper; base-combine assertions untouched.
The Moxie model set is MiniMax H3-only, so no other model choices are
exposed in the editor:
- Moxie Multimedia Loader: removed the clip_type dropdown (which listed
every CLIPType recipe in ComfyUI). The H3 Qwen3-VL text encoder is now
always loaded with CLIPType.MINIMAX; older ComfyUI builds raise a
clear error instead of silently loading the wrong encoder.
- Moxie Multimedia Loader: outputs are now exactly model, clip,
video_vae, audio_vae - the legacy FAST_MODEL_LOADER (model_loader)
bundle output is removed. The model output keeps the attention
backend and LoRA applied.
- Moxie MultiTrack Project: takes model, clip, video_vae and audio_vae
as required direct inputs instead of a bundled model_loader; the
fallback plumbing is removed while EasyMultiTrackProject keeps its
model_loader input for non-Moxie workflows.
- Moxie Preview Override: dropped the model_loader pass-through slots
and patches the model input directly.
- Moxie-consolidated.json reference workflow updated to the new slots.
- Added tests/test_moxie_nodes.py covering the schema contracts, the
attention -> LoRA -> output chain, the MINIMAX lock and the project
loader synthesis.