01
Greeting
Frontal photo of grandma and recording: «Happy birthday, grandson! Grow big and happy!»
Result: Talking greeting video
Try it →
BD
ByteDance · Avatars
New
OmniHuman 2.0
Photo talks and sings with your voice — new version
up to 30 s720p
22 gen / second of videoAvailable from plan Premium
Preview from the official model page on Atlas Cloud
No prompt needed
Pick an effect — upload your photo
No writing needed: tap an effect you like, add your photo, and in a minute your video is ready in the same style.
Overview
What it can do OmniHuman 2.0
OmniHuman 2.0 from ByteDance — new version of «talking photo»: send a face photo and voice recording, get a clip where the person speaks or sings, moving their head and hands. Length matches audio, up to 30 seconds.
Best suited for
- ✓Lips in sync with audio
- ✓Lively facial expressions and gestures
- ✓Up to 30 seconds
- ✓Lips synced precisely to audio: speech, song, rap
- ✓More natural head and hand movements than 1.5
- ✓Audio up to 30 seconds — the video will be just as long
How to write prompts
Formula for a good prompt for OmniHuman 2.0
- Photofront view, even lighting
- Textwhat to say
- Tonehow to say
- Lengthup to a minute
Tips specifically for this model
- Send a clear frontal face photo and voice audio. Price depends on audio length.
- Clear frontal photo and clean noise-free recording.
✕ Bad
say something about the product
✓ OK
Text: «Our service saves one hour a day — I will show how in one minute». Tone: friendly, confident.
Examples
Ready-made prompts for high quality results
Copy the prompt, insert your details — and get a clip matching the sample.
02
Tested in AskMeAI studio
Speaks calmly, smiles slightly, nods at the end of the phrase.
Result: the prompt used to test the model in the bot
Try it →
What not to do
Mistakes that ruin the result
In OmniHuman 2.0
- There must be only one person in the photo.
- Dark and blurry photos will not work.
- Multiple faces in the photo — only one person needed
- Dark and blurry shots
General rules for avatars
- Do not use another person’s photo or voice without their consent.
- Do not pick profile photos, sunglasses, or hands covering the face.
- Do not write text longer than one minute per clip.
Specifications
OmniHuman 2.0 in numbers
- Section
- Avatars
- Developer
- ByteDance
- Duration
- 2–30 s
- Quality
- 720p
- Sound
- no sound
- Price
- 22 gen / second of video
- Available from plan
- Premium
- Status
- works on the website and in the bot
Price
All plans →
How many costs
22 gen / second of video
| Plan | Price | Enough for |
|---|---|---|
| Free | free | — |
| Premium | 450 ★ · 1 month | up to 20 videos |
| Premium | 1 200 ★ · 3 months | up to 61 videos |
| Premium | 2 250 ★ · 6 months | up to 122 videos |
| Premium | 4 200 ★ · 12 months | up to 245 videos |
Questions about OmniHuman 2.0
Did not find an answer? Message us, real people reply.
Contact supportHow much does OmniHuman 2.0 cost?
22 generations per video second. Generations are the AskMeAI currency: included in Premium and available as one-time packs.
Can I try OmniHuman 2.0 for free?
OmniHuman 2.0 is available with Premium subscription and one-time generation packs. Try photo effects and chat with free models at no cost.
How to get the best result in OmniHuman 2.0?
Send a clear frontal face photo and voice audio. Price depends on audio length. Use the formula: photo → text → tone → length.
Where does OmniHuman 2.0 work?
On the askmeai.ru website studio and in the @GPT5Telegbot Telegram bot — shared balance.
Similar neural networks
All avatars →H
HeyGen VideoHeyGen · Avatars
New
Videos with dialogue and sound for creators and business
- Speech, noise, music in one pass
- Up to 15 seconds
up to 10 s768pbrings photos to life
24 gen / 5 s videoLearn more →
BD
OmniHuman 1.5ByteDance · Avatars
New
Your photo speaks or sings to your voice: lips, facial expressions, and gestures
- Lips in sync with audio
- Songs and speech up to 30 seconds
- Head and hands move
up to 30 s1080p
19 gen / second of videoLearn more →
H
HeyGen Avatar IVHeyGen · Avatars
TopSoon
Talking avatar: your photo reads any text with lifelike facial expressions
- Accurate lip sync
- Any voiceover language
- Gestures and emotions
39 gen / 5 s videoLearn more →