comfyui-mcp/packs/wan-animate/pack.yaml

Local-first, agent-native control plane for ComfyUI — MCP server + autonomous sidebar agent that drives your live graph in natural language on ANY LLM: Claude/ChatGPT/Gemini on your subscription (no API key), free local models via Ollama (fully offline),

name: wan-animate
display_name: WAN Animate 2.2 (v2v character replace)
family: wan
kind: video
description: >-
  WAN Animate 2.2 — video-to-video character / motion replacement built on the
  kijai WanVideoWrapper ecosystem. Drives a reference character with the pose,
  expression and motion of a driving video (ViTPose + YOLOv10 detection, SeC
  precise video segmentation for masking, relight LoRA for compositing). Uses the
  14B fp8-scaled WAN Animate model plus the lightx2v step-distill LoRA for fast
  sampling. For the maintained/recommended character-animation pipeline, prefer the
  `artokun-flow` pack.
vram: 24GB+
workflow: workflow.json
skill: wan-t2v-video
launch_args: []
sources:
  mirror: https://huggingface.co/Aitrepreneur/FLX
  wanvideo_wrapper: https://github.com/kijai/ComfyUI-WanVideoWrapper
  wan_animate_preprocess: https://github.com/kijai/ComfyUI-WanAnimatePreprocess
  secnodes: https://github.com/9nate-drake/Comfyui-SecNodes
notes:
  - Run the generated installer from your ComfyUI root (the folder containing
    custom_nodes/ and models/). On the Windows portable build that is
    ...\ComfyUI_windows_portable\ComfyUI\.
  - Requires SageAttention + Triton for the WanVideoWrapper sampler. Install them
    with install_triton_and_sageattention_auto.bat (run from
    ...\ComfyUI_windows_portable\python_embeded\). It detects your Torch/CUDA,
    installs triton-windows and the matching woct0rdho SageAttention wheel, and
    drops the Python include/libs zip so Triton can compile.
  - 'Dtype error "self and mat2 must have the same dtype": delete and re-clone
    kijai/ComfyUI-WanVideoWrapper, then reinstall its requirements.txt with the
    portable python (python_embeded\python.exe -m pip install -r
    ComfyUI-WanVideoWrapper\requirements.txt).'
  - 'RTX 5000-series onnxruntime "QuickGelu" / CUDA provider error on the ViTPose
    detection: pin onnxruntime==1.20.1 (pip install onnxruntime==1.20.1).'
  - '"Torch not compiled with CUDA enabled": your torch is a CPU build —
    reinstall torch/torchvision/torchaudio from the matching CUDA index, e.g.
    pip install torch torchvision torchaudio --index-url
    https://download.pytorch.org/whl/cu128 (use the cuXXX matching your driver).'
  - Masking has two paths — SeC precise masking (SeC-4B model, manual/prompted)
    or auto masking. The bundled SAM2 auto-download node
    (sam2.1_hiera_base_plus) is muted in the shipped workflow; SeC is the
    default. Pick the matching mask / background_image getter per the in-graph
    notes.
post_install:
  - Restart ComfyUI (or use ComfyUI-Manager -> Install missing custom nodes for
    any node Python deps).
  - Run install_triton_and_sageattention_auto.bat before first generate.
  - Load workflow.json.