Video Decomposer

This is the workflow I use to save time processing my videos. Sometimes, when you make adjustments to your workflow, the ControlNet processors might re-run videos that you have already processed. When you don't have a powerful GPU, saving time and memory becomes crucial. Pre-processing your ControlNets can save you both.

This workflow can save videos and images. It will return outputs for OpenPose, Canny, SoftEdge, LineArt, Depth, Main Subject Mask, Selective Mask with prompt, NormalMap, and the original video with the compatible frame rate and size. Additionally, this workflow can remove the background of the original video, restore faces, and perform face swaps.