OmniHuman 2.0 · Clipia

OmniHuman 2.0 a photo that talks and gestures

Upload a photo and a speech recording, and the person in the picture says your words, gestures and moves with the voice. The whole figure comes alive, not just the lips.

Create with OmniHuman 2.0 Watch examples

speech recording: 4–60 s
in the photo's proportions: 720p
face, hands, body: whole figure

In short

What is OmniHuman 2.0

OmniHuman 2.0 is a ByteDance talking-photo model. Upload one image and a 4 to 60 second speech recording: the model renders a 720p video in the photo's proportions where the lips match the words and the head, shoulders and hands move with the rhythm of the voice. An optional hint of up to 300 characters refines the manner.

photo + recording: all the input it needs
4–60 s: recording and clip length
up to 300 characters: motion hint
no watermark: on the finished clip

Photo + recording · product review

A creator reviews a serum

Ten seconds: a young woman holds a bottle up to the camera and calmly talks about it. The lips match the words, her other hand gestures, her head tilts in the pauses, and the bottle stays in her hand to the end.

How to repeat it

Use a photo where the product is already in hand and clearly visible: the model animates what is in the frame and does not add objects. Describe the manner in the hint: "talks calmly to the camera, slight smile".

Try it with your photo

Photo + recording · gestures

A storyteller in a library

An older man in a cardigan talks in front of bookshelves. His hands rise with the phrases, his shoulders and body follow the speech, and his features stay the same from the first frame to the last.

How to repeat it

For gestures you need a waist-up photo with the hands visible. Pauses and stresses in the recording become the rhythm of the movement, so speak the way you want the character to move.

Try it with your photo

3D character · emotions

A cartoon character celebrates

A 13-second emotional recording: the character is surprised, laughs and says thank you. The face follows the intonation, from wide-open eyes to a shy smile at the end.

How to repeat it

The model animates more than photos: an illustration or a 3D character with a clearly visible face works too. The voice sets the emotion, so the livelier the recording, the more expressive the face.

Try it with your photo

Open the model in the video studio. The editor shows the limits and the cost.

Create with OmniHuman 2.0

Parameter	Value	Note
Duration	4–60 seconds	The clip is as long as the recording; there is no separate length setting
Resolution	720p	No quality options
Mode	Talking photo	One photo and one speech recording per generation
Sound	Your recording	The model does not synthesize a voice: the clip plays the uploaded recording
Frame format	Photo proportions	A vertical photo gives a vertical clip, a horizontal one a horizontal clip
Inputs	Photo, speech recording, hint up to 300 characters	Recording in WAV, MP3, M4A, AAC, WEBM or OGG; the hint is optional

Photo and recording to video

Upload a picture and a 4 to 60 second speech recording: the person in the photo says it with synced lips and facial expressions.

Try it with your photo

Gestures and body movement

Hands, shoulders and body move with the speech, so a waist-up or full-length photo comes alive as a whole.

People, drawings and 3D characters

A photo, an illustration or a 3D character with a visible face all work. One photo may show several people.

Motion hint

Up to 300 characters on the manner: mood, gestures, gaze. Without a hint the model takes rhythm and emotion from the recording.

Parameter	OmniHuman 2.0	Kling Lip Sync
What moves	The whole figure: face, head, hands, body	Face and head
Which photo fits	Close-up, waist-up or full-length; several people allowed	A front-facing portrait with one face
Recording length	4–60 seconds	3–60 seconds

Pick a photo

A clear, unobstructed face. For gestures, a waist-up shot with the hands visible.

Add a speech recording

4 to 60 seconds of clean voice without music or noise: the recording length becomes the clip length.

Set the manner

Optionally describe the mood and gestures in a hint of up to 300 characters.

Prepare a photo and a speech recording.

Create with OmniHuman 2.0

A clip from a 4-second recording costs 29 credits, 10 seconds costs 72 credits, one minute costs 432 credits.

The price follows the length of the uploaded recording: the model has no quality or duration settings.

The recording decides the most

Pace, pauses and intonation turn into facial expressions and gestures. Clean speech without music or background noise gives the most accurate lips.

The hint is about manner, not plot

Describe how the character speaks: "talks calmly to the camera, slight smile, small hand gestures". English is the safest choice for the hint.

The frame defines the clip

The video shows what is in the photo, in its proportions. Need a vertical clip for Reels or Shorts? Use a vertical photo.

Product review by a host

A photo with the product in hand and a short script become a clip for a listing or social media, with no shoot.

A narrator for a course or a channel

A host explains the topic with gestures and facial expressions; all you need is a photo and a voice recording.

A brand mascot

A drawn or 3D character speaks in the brand's voice and changes emotion with the intonation.

Before a request, check that the model is available in the catalog and get a cost estimate for the recording length.

Clipia documentation

FAQ

Is OmniHuman 2.0 available in Clipia?

Yes. The model is open in the video studio, in the talking photo mode next to Kling Lip Sync. The buttons on this page lead straight to it.

How is OmniHuman 2.0 different from Kling Lip Sync?

Kling Lip Sync animates the face and head of a front-facing portrait, which is enough for a close-up. OmniHuman 2.0 moves the whole figure: hands, shoulders and body follow the speech, so a waist-up or full-length photo works, including one with several people.

How long can a clip be?

The clip is as long as the recording: 4 to 60 seconds. There is no separate duration setting.

How much does a generation cost?

A clip from a 4-second recording costs 29 credits, 10 seconds costs 72 credits, one minute costs 432 credits. The editor shows the exact amount for your recording before you start.

Are the demos on this page real generations?

Yes. The three clips were made with OmniHuman 2.0 in 720p with no editing, each from one photo and one recording. The portraits and the 3D character were generated by AI and the speech is synthesized: no real people appear in the clips.

Will OmniHuman 2.0 be in Clipia API?

Yes, when the model has the active status in the Clipia catalog. Check the parameters and the cost before creating a task.

Create a video

Bring your photo to life in OmniHuman 2.0

Upload a picture and a speech recording. The cost is visible before you start.

Create with OmniHuman 2.0

OmniHuman 2.0 demo

Watch the clip and explore its motion, detail, and camera work.

Watch demo

Open OmniHuman 2.0 in the editor

OmniHuman 2.0 — русская версия

OmniHuman 2.0 · Clipia

OmniHuman 2.0 a photo that talks and gestures

Upload a photo and a speech recording, and the person in the picture says your words, gestures and moves with the voice. The whole figure comes alive, not just the lips.

Create with OmniHuman 2.0 Watch examples

speech recording: 4–60 s
in the photo's proportions: 720p
face, hands, body: whole figure

In short

What is OmniHuman 2.0

photo + recording: all the input it needs
4–60 s: recording and clip length
up to 300 characters: motion hint
no watermark: on the finished clip

Photo + recording · product review

A creator reviews a serum

How to repeat it

Use a photo where the product is already in hand and clearly visible: the model animates what is in the frame and does not add objects. Describe the manner in the hint: "talks calmly to the camera, slight smile".

Try it with your photo

Photo + recording · gestures

A storyteller in a library

An older man in a cardigan talks in front of bookshelves. His hands rise with the phrases, his shoulders and body follow the speech, and his features stay the same from the first frame to the last.

How to repeat it

For gestures you need a waist-up photo with the hands visible. Pauses and stresses in the recording become the rhythm of the movement, so speak the way you want the character to move.

Try it with your photo

3D character · emotions

A cartoon character celebrates

A 13-second emotional recording: the character is surprised, laughs and says thank you. The face follows the intonation, from wide-open eyes to a shy smile at the end.

How to repeat it

The model animates more than photos: an illustration or a 3D character with a clearly visible face works too. The voice sets the emotion, so the livelier the recording, the more expressive the face.

Try it with your photo

Open the model in the video studio. The editor shows the limits and the cost.

Create with OmniHuman 2.0

Parameter	Value	Note
Duration	4–60 seconds	The clip is as long as the recording; there is no separate length setting
Resolution	720p	No quality options
Mode	Talking photo	One photo and one speech recording per generation
Sound	Your recording	The model does not synthesize a voice: the clip plays the uploaded recording
Frame format	Photo proportions	A vertical photo gives a vertical clip, a horizontal one a horizontal clip
Inputs	Photo, speech recording, hint up to 300 characters	Recording in WAV, MP3, M4A, AAC, WEBM or OGG; the hint is optional

Photo and recording to video

Upload a picture and a 4 to 60 second speech recording: the person in the photo says it with synced lips and facial expressions.

Try it with your photo

Gestures and body movement

Hands, shoulders and body move with the speech, so a waist-up or full-length photo comes alive as a whole.

People, drawings and 3D characters

A photo, an illustration or a 3D character with a visible face all work. One photo may show several people.

Motion hint

Up to 300 characters on the manner: mood, gestures, gaze. Without a hint the model takes rhythm and emotion from the recording.

Parameter	OmniHuman 2.0	Kling Lip Sync
What moves	The whole figure: face, head, hands, body	Face and head
Which photo fits	Close-up, waist-up or full-length; several people allowed	A front-facing portrait with one face
Recording length	4–60 seconds	3–60 seconds

Pick a photo

A clear, unobstructed face. For gestures, a waist-up shot with the hands visible.

Add a speech recording

4 to 60 seconds of clean voice without music or noise: the recording length becomes the clip length.

Set the manner

Optionally describe the mood and gestures in a hint of up to 300 characters.

Prepare a photo and a speech recording.

Create with OmniHuman 2.0

A clip from a 4-second recording costs 29 credits, 10 seconds costs 72 credits, one minute costs 432 credits.

The price follows the length of the uploaded recording: the model has no quality or duration settings.

The recording decides the most

Pace, pauses and intonation turn into facial expressions and gestures. Clean speech without music or background noise gives the most accurate lips.

The hint is about manner, not plot

Describe how the character speaks: "talks calmly to the camera, slight smile, small hand gestures". English is the safest choice for the hint.

The frame defines the clip

The video shows what is in the photo, in its proportions. Need a vertical clip for Reels or Shorts? Use a vertical photo.

Product review by a host

A photo with the product in hand and a short script become a clip for a listing or social media, with no shoot.

A narrator for a course or a channel

A host explains the topic with gestures and facial expressions; all you need is a photo and a voice recording.

A brand mascot

A drawn or 3D character speaks in the brand's voice and changes emotion with the intonation.

Before a request, check that the model is available in the catalog and get a cost estimate for the recording length.

Clipia documentation

FAQ

Is OmniHuman 2.0 available in Clipia?

Yes. The model is open in the video studio, in the talking photo mode next to Kling Lip Sync. The buttons on this page lead straight to it.

How is OmniHuman 2.0 different from Kling Lip Sync?

How long can a clip be?

The clip is as long as the recording: 4 to 60 seconds. There is no separate duration setting.

How much does a generation cost?

A clip from a 4-second recording costs 29 credits, 10 seconds costs 72 credits, one minute costs 432 credits. The editor shows the exact amount for your recording before you start.

Are the demos on this page real generations?

Will OmniHuman 2.0 be in Clipia API?

Yes, when the model has the active status in the Clipia catalog. Check the parameters and the cost before creating a task.

Create a video

Bring your photo to life in OmniHuman 2.0

Upload a picture and a speech recording. The cost is visible before you start.

Create with OmniHuman 2.0

OmniHuman 2.0 a photo that talks and gestures

What is OmniHuman 2.0

Three clips made with OmniHuman 2.0

A creator reviews a serum

A storyteller in a library

A cartoon character celebrates

OmniHuman 2.0 technical specifications

What OmniHuman 2.0 can do

Photo and recording to video

Gestures and body movement

People, drawings and 3D characters

Motion hint

OmniHuman 2.0 vs Kling Lip Sync

How to create a talking video in OmniHuman 2.0

Pick a photo

Add a speech recording

Set the manner

Pricing in Clipia

How to get a lively clip in OmniHuman 2.0

The recording decides the most

The hint is about manner, not plot

The frame defines the clip

Where OmniHuman 2.0 helps

Product review by a host

A narrator for a course or a channel

A brand mascot

Integration via Clipia API

OmniHuman 2.0 FAQ

Bring your photo to life in OmniHuman 2.0

OmniHuman 2.0 demo

Related Models

OmniHuman 2.0 a photo that talks and gestures

What is OmniHuman 2.0

Three clips made with OmniHuman 2.0

A creator reviews a serum

A storyteller in a library

A cartoon character celebrates

OmniHuman 2.0 technical specifications

What OmniHuman 2.0 can do

Photo and recording to video

Gestures and body movement

People, drawings and 3D characters

Motion hint

OmniHuman 2.0 vs Kling Lip Sync

How to create a talking video in OmniHuman 2.0

Pick a photo

Add a speech recording

Set the manner

Pricing in Clipia

How to get a lively clip in OmniHuman 2.0

The recording decides the most

The hint is about manner, not plot

The frame defines the clip

Where OmniHuman 2.0 helps

Product review by a host

A narrator for a course or a channel

A brand mascot

Integration via Clipia API

OmniHuman 2.0 FAQ

Bring your photo to life in OmniHuman 2.0