antgroup / echomimic

EchoMimic: Lifelike Audio-Driven Portrait Animations through Editable Landmark Conditioning
https://antgroup.github.io/ai/echomimic/
Apache License 2.0
3.01k stars 349 forks source link

About Conditions for Inference and Training #147

Open Shelton0215 opened 2 months ago

Shelton0215 commented 2 months ago

hello, I have a question, were the audio-driven and audio + pose-driven models trained separately? Is the weak condition for audio training audio and face mask?