A somewhat finished version of my ComfyUI Video2Video Frame inpainting workflow for Wan VACE. This is based on the ComfyUI First Frame Last Frame Wan Vace template. The changes I have made allow the use of videos instead of images for better context and motion of the inpainted frames.
What this is for.
This allows for the inpainting of frames between two video clips, to create longer videos or transitions between videos while maintaining better context and motion.
Tips for usage.
Some tips for using this workflow. It is a good idea to use videos with a similar context and that you describe the motion you want to happen between them. If the videos are wildly different or there just isn't enough frames for the motion you want, it will most likely just hard cut or crossfade to the the other clip.
The base setup for the workflow is made to use around 1 second of the standard 16fps Wan video output, and you will need to change the settings to limit the number of frames for your videos to 17 frames. You can change the number of input frames for each video, but remember that adding more frame will use more ram and take longer to process.
You can make sure your frames are lined up with the preview nodes.