OmniHuman
OmniHuman is a human video generation model from ByteDance. It brings a single image to life from an audio signal, generating speech, expressions and natural body movement. Per the OmniHuman project page, it handles portrait, half-body and full-body images. In ImagineArt it appears in Lipsync Studio as OmniHuman - Bytedance, described as “Realistic talking Avatars”.Specifications
How to use
1
Open Lipsync Studio
Go to Video and select Lipsync Studio from the Video Tools tabs.
2
Select the model
Open Select Model and choose OmniHuman - Bytedance.
3
Add your image, audio and prompt
Upload the image under Upload Image, then click Add Audio or Generate Audio. Use Prompt to describe gestures or mood.
4
Generate
Click Generate.