Video2Video in ComfyUI [AnimateDiff] - V

This is a test focused on IPAdapter regional conditioning: by using a total of 4 animated masks and assigning an IPAdapter image to each one of them I managed to precisely assign and control different style transfers to selected parts of the animation.

The workflow uses:

- RGB render of the animated character for latent image encoding (in 512px).

- 4x luma masks for IPAdapter regional conditioning.

- Redshift Toon Shader render assigned to Canny edge detector.

- AnimateDiff (TemporalDiff)

- Unsampling/Resampling technique both for the 512px generation and first step upscaling to 896px.

- 4xUltrasharp upscaling model from 896px to 1024px.

[impunx]

Amazing, would like to see the pipeline ! Love from India 🎉

[noahbalanda]

These are fantastic! Would definitely pay for access to your workflow or a tutorial. Amazing stuff.