Pytorch-Lightning==1.9.0 decord kornia timm open_clip_torch av omegaconf transformers einops scikit-learn taming-transformers-rom1504 pandas fairscale psutil==5.9.5 annotated-types==0.5.0 plotly imageio==2.14.1 imageio-ffmpeg==0.4.7 opencv-python==4.8.0.74 moviepy Pillow tqdm webdataset Fire natsort wandb clip @ git+https://github.com/openai/CLIP.git

ComfyUI/models/checkpointsbase workflow
https://github.com/chaojie/ComfyUI-MotionCtrl-SVD/blob/main/workflow.json
A little exploration workflow: for videos with relatively static camera angles, applying the same MotionCtrl to each frame, then combining images from corresponding positions to create a new video. The nth frame image is taken from the generated nth frame and merged again.
https://github.com/chaojie/ComfyUI-MotionCtrl-SVD/blob/main/workflow_video.json
original video: https://github.com/chaojie/ComfyUI-MotionCtrl-SVD/blob/main/assets/original.mp4
generate video: https://github.com/chaojie/ComfyUI-MotionCtrl-SVD/blob/main/assets/svd.mp4