Poor LipSync Quality at lower resolution.

BadToBest / EchoMimic

Lifelike Audio-Driven Portrait Animations through Editable Landmark Conditioning

https://badtobest.github.io/echomimic.html

Apache License 2.0

2.26k stars 263 forks source link

Closed 345ishaan closed 1 month ago

345ishaan commented 1 month ago

If i decrease the resolution to 256x256, there is no lip movement. I want to run at lower resolution for faster inference.

yuange250 commented 1 month ago

we only train the model on 512*512, so it works bad on low resolution :(