What this workflow does

• Create a video from the input image using Stable Video Diffusion
• Enhance the details with Hires.fix + video2video using AnimateDiff
• Fix the face with faces wap(ReActor) if it changes
• Finally apply frame interpolation(IFUNet) to finish👏

How to use this workflow

• Drag and drop your favorite image to the LoadImage node.
• Press the Queue Prompt button.

🚨Notice

• It seems that IFUNet.pth is currently unavailable for download, and there may be people encountering errors with the IFUnet VFI node.
• As there is no workaround, please enter “VFI” in the node search and try other VFI nodes such as FILM and RIFE.

Examples

• https://gyazo.com/51007c86e8b28b9ef7fc2e077c8e30b5
• https://gyazo.com/d9eac051c0697f49ed5b2e20d8dcea2f
• https://gyazo.com/eb810c03d40d3332865aae025934d8c2

Update

• 2023-12-21 : Codeformer/GFPGAN is now built into the ReActor node, so we simplified that part.