Omnihuman

Omnihuman

Omnihuman on FuseAITools is ByteDance's audio-driven digital-human video family. 1.5 Portrait Animation turns a single portrait and an audio track into a lip-synced talking head (audio under 60s, 720p/1080p, optional masks and seed); 2.0 Audio to Video animates one source image from audio clips of 4–300s. Both bill per second of audio duration—credits shown on Generate before submission. New users receive 10 free credits on sign-up.

Features

Core capabilities

1.5 Portrait Animation (omnihuman-1-5): Required **one portrait image URL** + **one audio URL** + **duration (1–59s, audio must be under 60s)**. Optional **up to 5 mask image URLs**, **prompt (≤300 chars)**, **output resolution 720 or 1080 (default 1080)**, **fast mode** and **seed**. Billed per second of audio duration.

2.0 Audio to Video (omnihuman-2-0): Required **one source image** + **one audio URL** + **duration (4–300s)**. Optional **prompt (≤300 chars)** and **watermark toggle (default on)**. Billed per second of audio duration.

Workflow

Upload a clean front-facing portrait (1.5) or source image (2.0), then add the voice audio you want lip-synced. The duration is read from the audio and drives pricing. For 1.5, optionally add region masks, choose 720p/1080p, enable fast mode for drafts, or lock a seed for reproducibility. Preview and download when the render completes.

Try Omnihuman on FuseAITools

Open 1.5 or 2.0 from the tabs, drop in an image and an audio track, and credits per second appear on the Generate button before you submit. No local GPU required. See pricing for member discounts and credit top-ups.