Fuser Apps are here 🚀
Fuser Apps are here. Free generations for the next month 💫
Let's GobyTencent Hunyuan 3D Digital Human Team
Generate physics-aligned 3D humanoid skeletal animations and export ready-to-rig motion capture sequences directly into your game engine
From natural language prompt to production-ready skeletal data in three straightforward steps.
Built on a billion-parameter Diffusion Transformer with Flow Matching and reinforcement learning, delivering physics-grounded skeletal kinetics without manual keyframing.
Explore generated motion captures spanning athletic acrobatics, combat choreography, subtle emotional gestures, and natural bipedal locomotion.
Game studios, VFX animators, and digital artists rely on text-to-motion generation to accelerate character animation pipelines and rapid pre-visualization.
Yes, HY-Motion 1.0 is the official technical designation for Hunyuan Motion, developed by Tencent's Hunyuan 3D Digital Human Team. It refers to the open-source, billion-parameter Diffusion Transformer model trained on over 3,000 hours of skeletal motion data.
The model outputs 3D motion animations exported natively as FBX files running at 30 frames per second. Output skeletons adhere to standard SMPL and SMPLH bone naming hierarchies, allowing seamless retargeting onto custom humanoid character meshes in Unreal Engine, Unity, Blender, or Maya.
Hunyuan Motion excels at generating solo bipedal humanoid locomotion, athletic sequences, combat choreography, and expressive physical gestures. For best results, structure prompts using an Action + Style + Emotion + Tempo formula that focuses strictly on skeletal kinematics rather than environmental surroundings.
The model is restricted to single humanoid actors and cannot generate quadrupedal animals, multi-character contact choreography, or physical interactions with external objects like climbing ladders. In addition, hand animations capture overall wrist and arm motion without localized finger articulation.
Duration can be set between 0.5 and 12 seconds—use 2 to 4 seconds for simple actions and 6 to 10 seconds for multi-stage sequences to prevent unnatural slow-motion or clipping. A guidance scale of 5 to 7 offers the best balance between prompt adherence and fluid physical momentum.
Generate physics-aligned 3D humanoid skeletal animations and export ready-to-rig motion capture sequences directly into your game engine