No such language.
The decision is in the “What it needs as input” row of the table. OmniHuman needs a PHOTO and audio — it makes a still frame speak. Sync needs a VIDEO and audio — it takes an already filmed clip and matches the lips to a new recording. No video, only a portrait — OmniHuman. A filmed person and new audio — Sync. The two are not interchangeable.
The price in credits is per SECOND of video and is shown on the “Generate” button before every run. The “Five seconds cost” row below is the same price multiplied by five.
| What | OmniHuman | Sync |
|---|---|---|
| Model variants on our platform | 1 | 1 |
| What it needs as input | photo, audio file | video, audio file |
| Price in credits per second | 321 | 167 |
| Five seconds cost | 1605 credits | 835 credits |
| Longest clip at once | 60 sec | 60 sec |
| Resolutions to choose from | 720p, 1080p | — |
| Sound from the model itself | ✗ | ✗ |
| Negative description (“without…”) | ✗ | ✗ |
OmniHuman is the only one here that starts from a still photo. A portrait, an archive frame or a generated face becomes a talking person, with no camera needed.
It has a choice of resolution — a rarity among talking-person models — and the option to supply a mask, meaning you can specify which part of the frame to animate.
The length it accepts is the same as with Sync. So not having a video as input doesn't cost you a shorter result.
Sync keeps all the movement from the original clip — gestures, posture, gaze — and changes only the lips. This is more convincing than any photo animation, because the rest is real.
Sync has several modes for when the audio and the clip aren't the same length: trimming, looping, reversing, silence.
It accepts a long clip — the same limit as OmniHuman.
Where OmniHuman loses. All the movement is invented by the model. The gestures and posture are only as convincing as the model made them, and in a long recording the monotony shows. OmniHuman is also among the pricier ones per second: see the price in the table.
Where Sync loses. It needs an already filmed clip. If you have no video, Sync simply doesn't take part, and none of its settings help. And it changes only the lips: if the recording has laughter or a pause, the body in the frame doesn't react to them.
You have a photo — OmniHuman. You have a video — Sync. That's the whole choice, and it's not a matter of preference.
If you have both, Sync usually gives a more convincing result, because most of the frame is real. The full cards are on the OmniHuman page and the Sync page.
Sync. OmniHuman starts at 321 credits per second, Sync — at 167 credits per second. The cheapest option for OmniHuman is OmniHuman (photo → speech), for Sync — Sync 2.0 Pro Lip-sync. The exact price is shown on the button before you start the generation.
No. Sync needs a video as input. For a photo that starts talking, use OmniHuman.
Sync, when you have a video — because the body, gaze and gestures are real and only the lips change. OmniHuman invents everything.
Yes, the input is a photo, no matter where it comes from. The content rules apply — the face of a real, recognizable person is not animated without their consent.
This page compares the two brands with each other. The full technical card for each — all variants, their strengths and weaknesses, typical tasks and real examples from the community — is on its own page.
Last updated:
