-
-
Notifications
You must be signed in to change notification settings - Fork 55
Description
Hi there,
Thank you so much for providing a method for lip-syncing. I’m looking to achieve a sharp and crisp lipsynced mouth before applying face swapping. I’m not a big fan of face swapping phase in the workflow because it tends to degrade the mouth performance in the model, making the speech appear less natural. It seems like the lip-sync node requires higher resolution for the mouth region—how can I achieve this?
I’ve tried using a 1080x1920 video where the person’s head and chest are visible, but the mouth still lacks sufficient resolution. It would be fantastic to find a solution, such as processing the lip-sync at 2x or 4x the original resolution and then merging it back into video downsized to avoid blurriness. While face swapping somewhat addresses this issue, the teeth aren't very active and appear less prominent. For example, when the model opens their mouth, after face swapping only the lips move with just a little bit of the teeth visible, making the teeth look less lively. However, before face swapping, everything looks great except for the blurriness.
It would be really great to solve this issue.
Thank you!