Hunyuan3D-2

Hunyuan3D-2 advances Tencent’s research in text-to-3D by improving both the realism and usability of generated assets. It produces meshes and textures with higher geometric accuracy and fewer artifacts, while ensuring the output is lightweight enough for downstream workflows in engines like Unity or Unreal. Compared to the first release, it generates faster and with more consistent quality, allowing artists to iterate on character models, props, or environments in minutes rather than hours.

The model supports prompts, sketches, or reference images as input, letting users control style and structure. Generated outputs come with cleaner topology, UV maps, and PBR-ready textures, so they can be dropped directly into production pipelines with minimal cleanup. It’s particularly suited for gaming, AR/VR, digital twins, and virtual commerce, where large volumes of assets must be created quickly and at scale. Safety and guardrails are built in, and the system can align outputs with brand or project requirements, making Hunyuan3D-2 a practical bridge between creative ideation and production-ready 3D content.