Sonic is a method about ' Shifting Focus to Global Audio Perception in Portrait Animation',you can use it in comfyUI
- some guys cuda must use cuda:0,so fix it. 修复有些人的电脑必须用cuda:0,否则会报错的错误。
- fix bf16 error,fix 12GVRAM maybe OOM when first run,fix MPS device error,修复bf16无法使用的错误,修复12GVram首次加载时容易OOM的问题,修复MAC的MPS支持。
In the ./ComfyUI/custom_node directory, run the following:
git clone https://github.com/smthemex/ComfyUI_Sonic.git
pip install -r requirements.txt
- 3.1.1 download checkpoints from google 从Google下载必须的模型,文件结构如下图
- 3.1.2 download openai/whisper-tiny
-- ComfyUI/models/sonic/
|-- audio2bucket.pth
|-- audio2token.pth
|-- unet.pth
|-- yoloface_v5m.pt
|-- whisper-tiny/
|--config.json
|--model.safetensors
|--preprocessor_config.json
|-- RIFE/
|--flownet.pkl
- 3.2 SVD checkpoints svd_xt.safetensors or svd_xt_1_1.safetensors
-- ComfyUI/models/checkpoints
├── svd_xt.safetensors or svd_xt_1_1.safetensors
- Replace 'frame number' with 'duration',you can use it to change 'infer audio seconds'. 使用duration替换frame number选项,用于控制输出音频的长度(单位为秒),注意因为实际对比长度是音频振幅数组,不是百分比精准;
- Fixed the bug of batch mismatch when the frame rate is not 25.修复帧率不是25时,batch不匹配的bug。
- Change the model loading to a monolithic SVD model, 模型加载改为单体SVD模型;
- Support output of non square images,OOM 支持非正方形图片的输出,容易OOM;
- image_size is used to control the minimum size of the output image. If OOM, please reduce this value ,image_size用于控制输出图片的最小尺寸,如果OOM请调小这个数值;
- 感谢@civen-cn 提交的PR
@article{ji2024sonic,
title={Sonic: Shifting Focus to Global Audio Perception in Portrait Animation},
author={Ji, Xiaozhong and Hu, Xiaobin and Xu, Zhihong and Zhu, Junwei and Lin, Chuming and He, Qingdong and Zhang, Jiangning and Luo, Donghao and Chen, Yi and Lin, Qin and others},
journal={arXiv preprint arXiv:2411.16331},
year={2024}
}
@article{ji2024realtalk,
title={Realtalk: Real-time and realistic audio-driven face generation with 3d facial prior-guided identity alignment network},
author={Ji, Xiaozhong and Lin, Chuming and Ding, Zhonggan and Tai, Ying and Zhu, Junwei and Hu, Xiaobin and Luo, Donghao and Ge, Yanhao and Wang, Chengjie},
journal={arXiv preprint arXiv:2406.18284},
year={2024}
}