[Request] VideoCoF WorkFLow: Current SOTA Open-Source Video Editing Model

[knightyxp]

Hi @kijai,

Could you look into supporting VideoCoF?

It is currently the state-of-the-art (SOTA) open-source video editing model. It uses a new "Chain-of-Frames" approach and supports fast 4-step inference.

Resources:

Repo: https://github.com/knightyxp/VideoCoF

Project Page: https://videocof.github.io/

HuggingFace Demo: https://huggingface.co/spaces/XiangpengYang/VideoCoF (You can try it here first!)

Would love to see this in ComfyUI! Thanks.

[RuneGjerde]

Looks interesting for sure ;-)
both style transfer and local editing

[knightyxp]

Looks interesting for sure ;-) both style transfer and local editing

Thanks! Yes — beyond style transfer and local editing, we also support instance-level editing including object addition, object removal, and object swap with only 4steps (very fast).
You can try it here: https://huggingface.co/spaces/XiangpengYang/VideoCoF