Skip to main content

ByteDance's OmniHuman 1.5 is the 1.5 model in its portrait-animation series, combining speech or singing with optional gesture and camera direction at 720p or 1080p.

From $0.144/s

Workbench

Input

Talking avatar
240/2000
Optional talking-avatar direction for action, expression, scene, and camera, limited to 2,000 characters. No default text is specified; the dialogue remains in the recording, and prompt text has no price effect. at least 1 characters · at most 2000 characters
1/1

Drop files or media links, paste, or browse · 1 file · Max 10 MB · JPG · PNG · WEBP

  • 1
1 / 1 files
Required for talking-avatar: JPEG, PNG, or WebP up to 30 MB, with a clear frontal face. Include the upper body and hands for gestures. No default; image choice does not affect the audio-second rate. at least 1 characters
affects price1/1
1 / 1 files
Required for talking-avatar: MP3 or WAV up to 15 MB and 1–60 seconds, supplying speech or singing. No default; its duration rounds upward for billing. ByteDance recommends 15 seconds or less for quality. at least 1 characters
For talking-avatar, choose 720p or 1080p; omitting the field selects 1080p. The two tiers carry the same rate, and measured output dimensions vary with the generated composition.
Seed
Optional talking-avatar integer from 0 through 2147483647; no default is specified. Fix it while changing one setting to help compare results, without an identical-output promise or any price effect. no lower than 0 · no higher than 2147483647
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
Example result

Missing the model you need?

BYTEDANCE · PORTRAIT PERFORMANCE

OmniHuman 1.5 API at a glance

Let the recording set the delivery, then direct a nod, a gesture, or the camera with text.

Audio accepted
1–60 s
ByteDance recommends staying within 15 seconds for quality; the recording sets the intended clip length.
Default output tier
1080p
720p is also selectable, with no change in the per-second rate.
Prompt character limit
2,000
Text can direct expression, movement, the scene, and the camera.
Portrait size ceiling
30 MB
Send JPEG, PNG, or WebP with a clearly visible frontal face.
Video delivery
MP4
Retrieve the completed file after an asynchronous task finishes.
Sample billable duration
7 s
The sleeper-train request supplied 6.35 seconds of audio, rounded upward for billing.

Results

What you can make with OmniHuman 1.5

  • INVITATION IN MOTION

    A palm-up gesture accompanies the line

    In our sleeper-train test, the still portrait became a speaking performance: she raised one hand palm up, nodded, and smiled. The 1080p request paired a 6.35-second recording with written movement direction.

    Measured 1080p test: a still portrait and 6.35-second recording returned 1248×1664 at 25 fps, with a raised palm, nod, and smile over 6.44 seconds.

  • CHECK THE COMPOSITION

    The portrait returned in a 3:4 frame

    The 1024×1536 train portrait produced a 1248×1664 video at 25 fps, showing the speaker from the waist up. Even with a prompt asking to keep the camera locked, the framing changed; inspect the result before placing it in an edit.

    Measured 1080p test: a still portrait and 6.35-second recording returned 1248×1664 at 25 fps, with a raised palm, nod, and smile over 6.44 seconds.

  • RECORDING PLUS DIRECTION

    Separate what is said from how it looks

    The audio file supplies the actual voice and dialogue. ByteDance says a prompt can guide gestures, mood, and camera movement while the recording drives the performance; this endpoint does not clone the speaker's voice.

  • TWO TIERS, ONE RATE

    Resolution is a picture choice

    The endpoint defaults to 1080p and also accepts 720p. Both use the same per-second price, calculated from the input recording and rounded upward rather than from the final video measurement.

Content test

Is OmniHuman 1.5 uncensored? What we tested

Content levels tested on OmniHuman 1.5, by mode
LevelTalking avatar
Mild
Not tested yet
Moderate
RendersTested Nov 15, 2026
Strong
Not tested yet
Very strong
Not tested yet
Maximum
Not tested yet
How we test

How each mode behaves

Talking avatar
Audio must fit 1–60 seconds and 15 MB in MP3 or WAV; ByteDance recommends 15 seconds or less for quality.
Talking avatar
A JPEG, PNG, or WebP portrait must be no larger than 30 MB; prompt text is limited to 2,000 characters.
Talking avatar
A resolution label does not lock the crop: our train sample changed from a 2:3 portrait to a 3:4 video.
Talking avatar
Gesture and camera instructions need visual review; the train sample confirms a raised hand and nod, without verifying every requested movement.
All modes
Generated files are kept for 14 days and then deleted; download any clip you intend to retain.
All modes
Use faces and voices you have the right to use; every person shown must be an adult, and no impersonation is allowed.

Use cases

What to build with the OmniHuman 1.5 API

  • Stage a short character invitation

    Use a portrait with room for the hands and a recorded invitation. The sleeper-train example demonstrates a raised palm, nod, and smile accompanying the supplied speech.

  • Record a presenter explanation

    Pair an authorized presenter portrait with a finished explanation and request a small nod or open-hand gesture. Keep the face clear and review the movement against the important words.

  • Animate a singing portrait

    Supply a sung recording instead of speech for the same photo-input task. ByteDance lists singing as a supported use; our documented runs here used spoken lines.

  • Try alternate acting directions

    Keep the portrait and recording steady while changing one expression or movement cue. Reuse a seed to help compare the change, then assess each video rather than expecting matching frames.

  • Prepare a portrait for a vertical edit

    Our two measured results retained vertical compositions, with different dimensions from their source images. Check the returned crop and visible gesture area before fitting a clip into your layout.

Prompts

OmniHuman 1.5 prompt examples you can copy

  • Talking avatar

    The tested train invitation

    Result from this prompt

    Begin with a gentle smile, lift one hand from the table in a small inviting gesture, then softly open both palms and give a playful nod. Keep the face frontal and clear, warm light steady, and camera locked in the original waist-up framing.

    Why it works. This exact direction accompanied our train example. One raised palm, the nod, and the smile were visible; the output changed to a 3:4 composition despite the locked-framing instruction.

    Try this prompt
  • Talking avatar

    A workshop explanation

    A furniture maker in his fifties wearing a cotton work shirt explains his process with a focused expression. He raises one open hand briefly during a pause, then lowers it. Hold a steady medium shot with his face visible.

    Why it works. Suggested direction for a portrait with visible hands: one gesture and a steady camera make the requested action easy to check.

    Try this prompt
  • Talking avatar

    A measured gallery response

    A curator in her forties wearing a dark tailored jacket listens for a beat, then speaks with a thoughtful expression and one small nod. Let the camera move slowly closer while keeping her face in view.

    Why it works. Suggested expression and camera cues use the prompt controls without supplying dialogue or claiming a tested camera result.

    Try this prompt
  • Talking avatar

    A relaxed window-side reply

    A woman in her late thirties wearing a silk robe sits beside a window and speaks with an amused smile. She tilts her head slightly as the line ends. Keep the camera still and the window light steady.

    Why it works. Suggested direction combines a single expression change with a small closing gesture for a source portrait in that setting.

    Try this prompt
  • Talking avatar

    A singer holds the mood

    A singer in his forties wearing an open shirt performs the supplied song with a reflective expression. His head rises slightly through the phrase and settles during the pause. Maintain the existing medium framing.

    Why it works. Suggested singing direction leaves the melody and voice in the audio. Singing is listed by ByteDance, while our supplied examples demonstrate speech.

    Try this prompt
  • Talking avatar

    A welcoming theater presenter

    A theater presenter in her fifties wearing a low-cut evening dress speaks with a welcoming smile. She opens one palm toward the viewer and gives a short nod. Keep the camera level and her face clear throughout.

    Why it works. Suggested prompt for a portrait with space around the hands; check the invitation gesture and crop in the completed video.

    Try this prompt

Pricing

OmniHuman 1.5 API pricing

A 5-second clip starts at $0.72.

Get started

How to use the OmniHuman 1.5 API in three steps

  1. Step 1

    Frame the face and gesture area

    Choose a frontal portrait with a clear face, including the upper body and hands if you want gestures. Use JPEG, PNG, or WebP within 30 MB and confirm rights to the likeness.

  2. Step 2

    Complete the audio performance

    Prepare MP3 or WAV within 15 MB and 1–60 seconds, using a voice you have permission to use. ByteDance recommends keeping recordings within 15 seconds for quality.

  3. Step 3

    Add direction and select resolution

    Describe visible movement or camera behavior in up to 2,000 prompt characters; the dialogue belongs in the recording. Choose 720p or the default 1080p at the same rate, with an optional seed for comparisons.

Parameter reference and code samples are in the API and Readme tabs at the top of this page.

FAQ

OmniHuman 1.5 API FAQ

About OmniHuman 1.5

Build the shot around the recording you intend to keep, then choose a portrait that leaves room for the requested movement. In our train sample, a short invitation brought a raised palm, a nod, and a smile into the still scene, but the composition changed to 3:4. Treat that crop as part of the review alongside the lip sync. Both 720p and 1080p use the same audio-second rate; trim the recording before submitting, check the returned MP4, and save it within 14 days. Use the related model pages when another set of inputs or limits suits the project.

Related models and next tasks