This is a ComfyUI custom node implementation of StableAnimator, allowing you to generate high-quality identity-preserving human image animations directly in ComfyUI.
- Clone this repository into your ComfyUI custom_nodes folder:
cd ComfyUI/custom_nodes
git clone https://github.com/your-username/StableAnimator .- Install the required dependencies:
pip install -r requirements.txt- Download the model weights and place them in the checkpoints folder:
- stable-video-diffusion-img2vid-xt
- Animation/pose_net.pth
- Animation/face_encoder.pth
- Animation/unet.pth
- Load the StableAnimator pipeline using the "Load StableAnimator Pipeline" node
- Connect your reference image and pose directory to the "Generate Animation" node
- Configure the generation parameters as needed
- Run the workflow to generate your animation
- width/height: Supported resolutions are 512x512 and 576x1024
- guidance_scale: Controls how closely to follow the pose guidance (recommended: 3.0)
- num_inference_steps: Number of denoising steps (recommended: 25)
- fps: Output video frame rate
- frames_overlap: Overlap between processed frame chunks (recommended: 4)
- tile_size: Size of processed tiles (recommended: 16)
- noise_aug_strength: Strength of noise augmentation (recommended: 0.02)
- seed: Random seed for reproducibility (-1 for random)
This project is licensed under the MIT License. See the LICENSE file for details.