ComfyUI Node
ONNX Detection Model Loader (V2)
Load ONNX ViTPose + YOLO detection models for Wan 2.2 Animate preprocessing. Place model files in `ComfyUI/models/detection/`. Outputs a `POSEMODEL` bundle that the detection node consumes.
ONNX Detection Model Loader (V2)
- model
◄vitpose_model(place .onnx models in ComfyUI/models/detection)►
◄yolo_model(place .onnx models in ComfyUI/models/detection)►
◄onnx_deviceCUDAExecutionProvider►
CategoryWanAnimatePreprocess_V2
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| vitpose_model | COMBO | (place .onnx models in ComfyUI/models/detection) | ViTPose ONNX file (human wholebody, e.g. vitpose_h_wholebody_model.onnx). Place in ComfyUI/models/detection/. .onnx is always listed here even if ComfyUI hides it elsewhere. |
| yolo_model | COMBO | (place .onnx models in ComfyUI/models/detection) | YOLO person-detector ONNX file (e.g. yolov10m.onnx — NOT a pose model). Place in ComfyUI/models/detection/. .onnx always listed. |
| onnx_device | COMBO | CUDAExecutionProvider | Execution provider for ONNX Runtime. CUDA is much faster; CPU is the safe fallback. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model | POSEMODEL | ViTPose+YOLO model bundle. Connect to `model` on Pose and Face Detection (V2). |