Models

AI models for 3D human motion.

Generate editable character animation from video, natural language, or explicit locomotion controls. Auto-rig a humanoid character, apply the motion, and export it as FBX, GLB, or BVH.

INPUTS
2D video
natural language prompt
direction · speed · stride · style
unrigged humanoid character
UTHANA MOTION MODELS
temporally consistent 3D skeletal motion
OUTPUTS
FBX · GLB · BVH
joints · rotations · root motion
editable keyframes

Choose the right model

Video-to-Motion

Convert a single-person video into editable 3D skeletal motion. Use the result for character animation or as the starting point for a structured motion-data workflow.

Explore Video-to-Motion

Text-to-Motion

Describe an action in natural language and generate full-body 3D character animation. Use it to explore an idea, create a starting point, or fill a motion gap without recording a performance.

Explore Text-to-Motion

Locomotion

Generate consistent, loopable travel motion with control over direction, speed, stride count, and movement style. Build related clips without relying on an open-ended text prompt.

Explore Locomotion

Auto-rigging

Turn a compatible, unrigged bipedal humanoid mesh into a rigged character ready for Uthana motion. Use it in the web product or through the GraphQL API.

Explore Auto-rigging

One motion pipeline

Start with an existing rig, or auto-rig a compatible humanoid character. Generate motion from a reference video, a natural-language prompt, or explicit locomotion controls. Apply the result to the target character with retargeting, then export it for the rest of the production pipeline.

Uthana returns skeletal animation rather than a rendered video. The motion remains available for retargeting, editing, and engine or DCC workflows after generation.

Build with the models

Use Uthana in the browser, or integrate the same motion workflows through the GraphQL API.