ByteDance has launched OmniHuman 2.0, an AI model that generates videos up to 5 minutes long from a single image and an audio recording.
ByteDance has launched OmniHuman 2.0, an AI model that generates videos up to 5 minutes long from a single image and an audio recording.
The model analyzes speech rhythm, tone, and meaning to animate realistic lip movements, facial expressions, gestures, and singing performances. Available through BytePlus, it supports 720p video generation at $0.07 per second.