Mixer AI
OmniHuman 2.0
A talking avatar from one photo and your audio recording: 720p video as long as the audio.
When to choose it
OmniHuman 2.0 brings one character photo to life: the character speaks and moves to the track you upload. You need exactly one photo and one audio file; a scene description is optional. Resolution and duration are not chosen: the result is always 720p and as long as the audio, and every second of audio is billed.
- A voice file instead of written lines: your recording sets the words.
- The video is as long as the audio, from 4 seconds to 5 minutes.
- An optional description of the scene, motion and camera up to 300 characters.
Modes and settings
- One photo (up to 30 MB, each side 300 to 6000 pixels, aspect ratio 0.4 to 2.5) and one MP3 or WAV audio file up to 15 MB.
- Audio from 4 seconds to 5 minutes. On the site and in Canvas other formats are converted to MP3; Telegram accepts only MP3 and WAV.
- The result is 720p; the photo sets the frame format and the audio sets the duration.
Price in coins
| Quality / size | Input | Coins / second |
|---|---|---|
| 720p | Portrait + audio | — |
Billing is per second, not per fixed-length clip. The rate is multiplied by the billable duration; the complete request total is rounded up to a whole coin once.
Duration follows the uploaded audio track: every second of audio is billed. The photo and the scene description add no seconds.
≈ marks a fractional rate rounded to two decimal places for display; billing uses full precision.
The dollar price follows the current coin rate.
What you can make
- A presenter for a training or ad video from a finished voice-over.
- A character delivering a line or a verse to an uploaded track.
How to direct the result
- Describe the scene, action and camera in one or two short phrases and do not repeat the voice-over text.
- Pick a photo with a clearly visible face and torso, without strong distortion or overlaps.
Example: a teacher gives a lesson
Photo: a waist-up portrait of a teacher with a well-lit face. Audio: a two-minute lesson recording without music. Description: calm light, medium shot, light gestures, almost static camera.
What to keep in mind
- Exactly one photo and one audio file are accepted; audio shorter than 4 seconds or longer than 5 minutes is rejected before billing.
- Resolution, frame format and duration cannot be chosen; a long track costs proportionally more.
Sources
Guides and comparisons using this model
FAQ
What do I need to upload?
One character photo and one MP3 or WAV audio file. A scene description is optional.
How long can the video be?
As long as the audio: from 4 seconds to 5 minutes.
Can I choose the resolution or the frame format?
No. The result is always 720p and the photo sets the frame format.
How is the price calculated?
Per second of uploaded audio at the rate in the table. The photo and the description add no seconds.
Will a Telegram voice message work?
Telegram accepts only MP3 and WAV. On the site and in Canvas other formats are converted to MP3 automatically.
Which photo should I choose?
One with a clearly visible face and torso and no strong overlaps; each side from 300 to 6000 pixels.
What changes the price?
The coin rate per second depends on quality, mode and, for some models, sound. Multiply it by the billable duration. Where input video is charged, its duration also counts. The complete request total is rounded once.
Do I need a separate model subscription?
No. In Mixer AI, you top up one coin balance and use it for the tasks you need. You do not need a separate subscription to this model.