ByteDance · video

OmniHuman 1.5

A photo and a voice track become a person talking and gesturing in sync, up to a minute long.

Talking photo
Resolution
720p, 1080p
References
Up to 2 image or audio files

About OmniHuman 1.5

OmniHuman 1.5 turns a photo of a person and a voice track into a video of that person speaking. According to the provider, the character's emotions and movements follow the audio closely. It runs on ByteDance's OmniHuman, priced at $0.40 per second of audio.

The newer OmniHuman 2 takes much longer audio at a lower rate, $0.27 per second. A typical generation takes about 5 min.

Good for

  • A person who talks and gestures with the voice track, not only moves the lips
  • The longest talking clips this model takes, at 720p

Getting good results

  1. Stay at 720p for long audio: the provider accepts a shorter track at 1080p, and MegaScenes refuses a track that is too long before any charge.
  2. The provider notes that its 720p generation is faster and higher in quality than 1080p; the price is the same at both.

OmniHuman 1.5 questions

How is OmniHuman 1.5 priced?

By the seconds of audio: $4.00 for a short line, $24.00 for a full minute. The resolution does not change the price.

Other video models

AI video generator