Hunyuan Motion

Generate 3D Character Animations with AI in Seconds. Transform natural language descriptions into professional 3D character animations — 50x faster than traditional motion capture.

View on GitHub Try Demo

Advanced Capabilities

Text-to-3D Motion Generation

Explore the powerful features that make Hunyuan Motion the leading open-source AI motion generation solution.

Text to Motion AI: Describe any motion in natural language — Hunyuan Motion converts your text prompt into realistic 3D skeleton animations.
Bilingual Support: Full support for both Chinese and English text prompts. The dual-encoder architecture understands complex motion descriptions in both languages.
SMPL/FBX Export: Export generated motions in SMPL/SMPLH skeleton format. Compatible with FBX for seamless integration with Blender, Unity, and Unreal Engine.

Simple Workflow

From Text to 3D Motion in 3 Steps

Describe Your Motion

Enter a text prompt like "a person walking confidently" or "performing a karate kick".

AI Generates 3D Motion

Hunyuan Motion's AI engine creates a realistic, physically accurate 3D skeleton animation in seconds.

Export & Use

Download the animation as BVH or FBX for use in Blender, Unity, or Unreal Engine.

Under the Hood

Technical Specifications

Model Architecture

ArchitectureDiffusion Transformer (DiT)
Sampling MethodFlow Matching
Text EncoderCLIP + Qwen3-8B
Output FormatSMPL/SMPLH Skeleton

Model Stats

Parameters (Full)1 Billion
Parameters (Lite)460 Million
Training Data3,000+ Hours
Motion Categories200+

Frequently Asked Questions about Hunyuan Motion

Hunyuan Motion is Tencent's open-source billion-parameter text-to-3D motion generation AI model. It allows you to generate professional 3D character animations from natural language descriptions in seconds.

Yes, Hunyuan Motion is free and open-source. You can access the model weights on Hugging Face and the code on GitHub. It is also available for commercial use under its license terms.

Hunyuan Motion supports exporting generated motions in SMPL/SMPLH skeleton formats. It is also compatible with FBX, making it easy to integrate with popular 3D tools like Blender, Unity, and Unreal Engine.

Yes, Hunyuan Motion features a dual-encoder architecture that supports both English and Chinese text prompts, understanding complex motion descriptions in both languages.

For the full 1B parameter model, an NVIDIA GPU with at least 24GB VRAM (like an RTX 4090 or A100) is recommended. The lite version (460M parameters) can run on GPUs with 14GB+ VRAM.