PicX Studio

VIDEOByteDance · from 40 credits / sec

Longer talking avatars from ByteDance

OmniHuman 2.0 Talking Avatar Generator

Photo + audio in. 720p. Audio from 4 seconds to 5 minutes.

OmniHuman 2.0 turns one photo and an audio track of up to five minutes into a lip-synced talking video at 720p. The video runs as long as your audio, and PicX turns kie's watermark off.

OmniHuman 2.0 output: photo + a 4-second voice line
OmniHuman 2.0 output: photo + a 4-second voice line
  • 720p
  • Video + audio

Long-form talking video

OmniHuman 2.0 output: photo + a 4-second voice line

Up to five minutes

OmniHuman 2.0 takes audio from 4 seconds to 5 minutes, enough for a full lesson, update or podcast segment in one clip.

The input photo (made with Seedream 5.5)

Lip sync with natural motion

Lips, face and head move with the speech, and an optional prompt of up to 300 characters can set the scene, the motion and the camera.

Billed by your audio

The price is per second of the audio you upload, measured before anything is charged, so a short line costs a fraction of a long talk.

OmniHuman 2.0, from its maker

Official release video by ByteDance

Official release video by ByteDance: Introducing OmniHuman 2.0 | Smarter Lip Sync, Longer Videos, Lower Costs on BytePlus. Not made on PicX Studio.
OmniHuman 2.0 output: the same voice on a new photo

OmniHuman 2.0

Photo + audio in. 720p. Audio from 4 seconds to 5 minutes.

Made for the work you ship

  • OmniHuman 2.0 output: photo + a 4-second voice line

    Lessons and training Long narrated clips with a presenter on screen.

  • Podcast and news clips A face for an audio show or a news read.

  • OmniHuman 2.0 output: the same voice on a new photo

    Characters Animated storytellers from an illustration.

  • Product walkthroughs Longer explainers voiced by one presenter.

OmniHuman 2.0

Why pick OmniHuman 2.0

Try OmniHuman 2.0
  • 5 minLong audioFrom 4 seconds to 5 minutes.
  • 720p720p outputOne resolution, the photo's shape.
  • Lower costCheaper per secondThan OmniHuman 1.5.
  • PromptScene and cameraOptional, up to 300 characters.
  • Per secondBilled by the audioMeasured before you are charged.
  • No markClean outputPicX turns the watermark off.

What OmniHuman 2.0 costs

Billed per second of video, at the resolution you pick. The same prices the generator shows before you run it.

  • 720p
    40credits / sec
    5s clip 200 creditsOn Starter≈ 15 clips / mo

Every model on every planCredits never expireSee plans

Pay for OmniHuman 2.0 with crypto

Buy credits with Bitcoin, USDT, USDC, ETH, Solana and more — no card, nothing auto-renews, packs from $5, and 5% extra credits on crypto payments.

How crypto checkout works

How to use OmniHuman 2.0 in 3 steps

  1. 1

    Upload a photo

    One clear photo or illustration with the face visible.

  2. 2

    Add a WAV or MP3

    Between 4 seconds and 5 minutes long, up to 10 MB (an MP3 fits five minutes easily).

  3. 3

    Add a prompt if you like, then generate

    Describe the scene or the motion in up to 300 characters, or leave it to the audio.

OmniHuman 2.0 prompts to copy

Real prompts and what they made. Copy one, or open it in the generator with the model already picked.

  • OmniHuman 2.0 output: photo + a 4-second voice line

    Studio welcome

    She talks to the camera with a friendly smile

    Try it
  • OmniHuman 2.0 output: the same voice on a new photo

    Same voice, new photo

    She talks to the camera on a rainy street at night, friendly and relaxed

    Try it
Browse all templates

OmniHuman 2.0 questions

What is OmniHuman 2.0?

OmniHuman 2.0 is ByteDance's newer audio-driven avatar model. It turns one image and an audio track into a realistic, lip-synced talking video at 720p.

What is new in OmniHuman 2.0 compared to 1.5?

OmniHuman 2.0 accepts much longer audio, up to five minutes instead of under one, and costs less per second, at 720p. OmniHuman 1.5 stays the pick for 1080p and short, expressive clips.

How can I use OmniHuman 2.0 online?

Open the generator from this page, upload one photo and your audio, add an optional prompt for the scene, and generate. The video is as long as your audio.

Is OmniHuman 2.0 free?

No. OmniHuman 2.0 is a paid video model on PicX Studio, billed per second of your audio. Credits never expire and work on every model.

What image and audio does OmniHuman 2.0 need?

One photo or illustration, and one WAV or MP3 file between 4 seconds and 5 minutes long, up to 10 MB on PicX.

Does OmniHuman 2.0 add a watermark?

Not on PicX: the model's watermark option is turned off for every generation.

Make long-form avatars with OmniHuman 2.0

Up to five minutes of speech on one photo, at 720p.