Generate 3D motion from text.

Describe an action in natural language. Uthana generates full-body character animation you can preview, retarget, and export as FBX, GLB, or BVH.

From description to editable animation

Text-to-Motion is for moments when you know what a character should do but do not have a reference video or animation ready. Write the action in ordinary language, generate a result, and preview it on a character.

Revise the prompt or generate another result to explore a different interpretation. Once the motion is right, apply it to a compatible bipedal character and continue editing it in the rest of the production workflow.

The output is skeletal animation, not a rendered video. That makes it useful as a starting point for production rather than a visual reference that must be recreated from scratch.

Use language for intent; use controls for locomotion

A text prompt is useful when the action, performance, or idea is easier to describe than to parameterize. It is not the right interface when exact travel direction, speed, stride count, and repeatability are the priority.

For predictable travel motion, use the Locomotion model. It exposes explicit controls for direction, speed, stride count, and movement style instead of relying on an open-ended prompt.

Apply the motion to your character

Use a built-in character or upload your own. If the character is not rigged, Auto-rigging can create a compatible humanoid skeleton. Uthana retargeting then adapts the generated motion to the target character’s skeleton and proportions.

Export and keep editing

Download the result as FBX, GLB, or BVH. Choose 24, 30, or 60 fps for the download. Continue editing the skeletal animation in the DCC or engine used by the production team.

Current model versions and model-specific generation parameters are documented in the API reference.

Start with a prompt

Describe the action, preview the generated motion, and apply it to a character.