ComfyUI_LatentSync

ComfyUI_LatentSync
★ 17

潜空间同步视频处理ComfyUI 插件依赖 ffmpeg
为 ComfyUI 提供非官方 LatentSync 实现，便于在潜空间中对视频帧或序列进行同步、编辑与合成。
💡 在 ComfyUI 内同步并编辑视频帧的潜空间信息
🍴 6 Forks💻 Python🔄 2025-01-06
📦 网盘链接待填入
📦 requirements.txt
torch
torchvision
--extra-index-url
https://download.pytorch.org/whl/cu121
xformers
--extra-index-url
https://download.pytorch.org/whl/cu121
triton
diffusers
transformers
huggingface-hub
imageio
decord
accelerate
einops
omegaconf
safetensors
opencv-python
mediapipe
av
torch-fidelity
torchmetrics
python_speech_features
librosa
scenedetect
ffmpeg-python
lpips
face-alignment
ninja
pandas
numpy
📄 README

ComfyUI LatentSync (Unofficial Support)

Unofficial LatentSync support for ComfyUI.

Important Updates

  • 2024/12/31: Added ComfyUI nodes and workflow examples

    • Example Workflow

    This workflow shows the basic usage.

    Installation

  • Install from ComfyUI Manager (search for latentsync, make sure ffmpeg is installed)

  • Download or git clone this repository into the ComfyUI/custom_nodes/ directory and run:
    • sudo apt install ffmpeg
sudo apt -y install libgl1
pip install -r requirements.txt

    Download Checkpoints

    All models will be downloaded automatically, if not, try download from HuggingFace repo, the checkpoints should appear as follows:

    ./checkpoints/
|-- latentsync_unet.pt
|-- whisper
|   `-- tiny.pt