search

MyNiuuu / MOFA-Video

[ECCV 2024] MOFA-Video: Controllable Image Animation via Generative Motion Field Adaptions in Frozen Image-to-Video Diffusion Model.
https://myniuuu.github.io/MOFA_Video
Other
615 stars 34 forks source link

face mask of facial landmark based animation. #50

Closed Jeoyal closed 1 month ago

Jeoyal commented 2 months ago

Dear Author, While reading the keypoint inference code, I noticed that there is no mention of the face mask component. Could this omission significantly impact the results? Have you considered incorporating this aspect, such as using the landmark range as a mask?

MyNiuuu commented 2 months ago

Does the term 'face mask' refer to the landmark mask utilized in hybrid control? If that's the case, I think it won't significantly affect the results in most situations. The mask in hybrid control is primarily used to restrict the landmark control area. Without it, the control area would be the entire image.