Using input video instead of input image

OpenTalker / SadTalker

[CVPR 2023] SadTalker：Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation

https://sadtalker.github.io/

Other

11.62k stars 2.17k forks source link

Using input video instead of input image #459

Open alexrichardson21 opened 1 year ago

alexrichardson21 commented 1 year ago

Is there a way to use an input video and have it just generate the new lip movements but leave the rest of the video untouched? I tried using --source_image as an mp4 file but it only uses the first frame as reference. wav2lip augments a whole video (at super low res) not just a single image and Sadtalker is built ontop of it so I feel like their must be a way to accomplish this

oijoijcoiejoijce commented 12 months ago

Lidong1211 commented 12 months ago

video-retalking SadTalker-Video-Lip-Sync