Open MetaInsight7 opened 1 month ago
I think in Advanced Liveportrait here's a way for you to use an image as a driver, plus nodes to modify the expression further when needed.
You can also use a program like Capcut to turn the images into short videos (basically the same image every frame for a few frames), use the vid to vid method, then extract the resulting picture with something like VLC ("video" pulldown then "take snapshot". But Advanced Liveportrait might be easier and better since there are nodes to refine the results if needed.
Thank you. I found that the advantage of LivePortraitKJ is that it decouples each module and can perform additional operations in each step. Unfortunately, it can only adapt to videos. Although Advanced Liveportrait can operate on pictures, it is not decoupled, so it is difficult to perform some additional operations.
I want to use images to drive images, but I found that there is no effect in the end. Does it not support inputting a single image