Workroom offers several lipsync models to sync a voiceover to a face. The default, Lipsync Labs, takes a video with a talking head. InfiniteTalk is different — it takes a still image and generates multi-person conversational video from it. Lipsync 2 Pro works like Lipsync Labs but produces higher output quality, up to 4K.
Model overview
| Model | Input | Best for |
|---|---|---|
| Lipsync Labs | Video + voiceover | General use, fast turnaround |
| InfiniteTalk | Image + voiceover | Multi-person conversational video generation |
| Lipsync 2 Pro | Video + voiceover | Higher output quality, up to 4K |
InfiniteTalk — image input
InfiniteTalk is the only model that accepts a static image instead of a video. It generates multi-person conversational video from a still photo — useful when you have a reference image but no recorded footage.
When you select InfiniteTalk, the Video button in the control bar changes to Image.
Switch your model
Open the model selector in the top-left corner of the Lipsync page. Selecting a model immediately updates the input controls in the bottom bar.