(For Patreon Supporters) Wan Vace - Multi Condition With Face Refiner Ver 20250611 (Patreon)
Videos
-
481075919.mp4
Downloads
Content
Let's make it short in the post, and I am focus on the additional features.
This workflow are based on The tutorial video : https://www.youtube.com/watch?v=XTYwSht6n9Y
And Wan Vace - Multi Condition Basic workflow : https://www.patreon.com/posts/comfyui-wan-2-1-131089902
Additional Feature, to add more detail for character facial expression.
Added FaceRefine group that is focus on adding facial expression, Ideally for medium or wider shot video. Close up shot, usually the first sampling group are good enough to render detail face. In diffusion models, Character usually starting to get less face detail from mid range.
For this group, you have 2 option for reference image.
1- use the reference image from the beginning of this workflow , point from the GetNode "img_ref_1"
2- Use another close up portrait image of your character. (You can use other face, for faceswap)
Connect either one to the Reroute node for Vace.
It will crop the face region , and use the source video as face pose reference. Then we have better facial expression for the AI character.
Still testing, and see if anything to improve , I will update newer version of it in the future.
Update - A - 20250611
I added another workflow file, that running the FaceRefiner in a separate workflow. (vace-native-v2v-Faceswap_FaceRefiner-gguf.zip)
You can load other video and try with the same face or different face as Faceswap feature.
update: For the face detection, you need this : https://github.com/kijai/ComfyUI-FollowYourEmojiWrapper
Files