OmniHuman 2.0

This site gives creators a simple browser studio for OmniHuman 2.0, ByteDance's audio-driven digital human model offered through BytePlus. It describes itself as an independent tool built on the model rather than an official ByteDance product. What makes the underlying model special is that audio leads the performance, so lip movement, intonation and emotion follow the sound and even the meaning of what is said.

• A single run can produce a 720p video up to five minutes long, a big jump from the 30 second limit of OmniHuman 1.5.
• Audio comprehension lets the character act on what the speech describes, rather than simply moving its mouth in time.
• Optional prompts direct chained actions, camera movement and background, giving more control than a plain lip-sync tool.
• It accepts JPG, PNG or WebP images and WAV, MP3 or M4A audio, which makes it easy to work with existing assets.
• New users start with free credits, and billing is per second of output, so short tests stay inexpensive.

It is well suited to musicians making performance videos, marketers building presenter clips, and educators producing explainers or motion comics. For the most convincing results, use a clear front-facing character image and clean audio without heavy background noise, then add a short prompt describing gestures or setting. The site's prompt guide and its comparison of versions 2.0 and 1.5 are worth a quick read before your first generation.