Tencent HY-World-2.0 is now public
The open-source model can reconstruct and simulate entire 3D environments from simple prompts.
Tencent has publicly released HY-World 2.0, a powerful multi-modal world model designed for 3D content creation. The model, now available on Hugging Face and GitHub, can reconstruct, generate, and simulate entire 3D environments from simple text or image prompts. This represents a major advancement in making sophisticated 3D world-building technology accessible to a broader developer community, moving beyond proprietary systems.
HY-World 2.0 demonstrates impressive capabilities in generating coherent and detailed 3D scenes, such as dense forests and structured urban landscapes, as shown in its online demo. The release includes comprehensive documentation and a technical paper, providing researchers and developers with the tools to experiment with and build upon the technology. This open-source approach could accelerate innovation in gaming, virtual production, and simulation.
- Open-source multi-modal model for 3D world generation and simulation
- Accepts both text and image inputs to reconstruct complex environments
- Publicly available on Hugging Face and GitHub with live demo
Why It Matters
Democratizes advanced 3D content creation for game developers, filmmakers, and simulation engineers.