Tencent open-sources HY-World 2.0 to advance interactive 3D content creation

New AI model enables developers to generate editable 3D environments from text, images, and video inputs, supporting gaming, simulations, and digital twin applications with seamless integration into popular engines

Tencent has unveiled and open-sourced its latest artificial intelligence model, HY-World 2.0, marking a significant advancement in 3D content generation. The model moves beyond traditional AI systems that create static images or videos, enabling the development of fully interactive and editable three-dimensional environments.

The release reflects a broader industry shift toward immersive digital experiences, with HY-World 2.0 designed to simplify and accelerate the creation of complex virtual worlds. By lowering technical barriers, the model is expected to benefit developers across gaming, simulation, and emerging digital twin use cases.

From prompts to playable worlds

A key highlight of HY-World 2.0 is its multi-modal capability, allowing users to input text, images, or video to generate 3D environments. Unlike earlier models that primarily produced visual outputs, this system creates structured 3D assets that can be modified and reused. These outputs combine formats such as 3D Gaussian splatting, meshes, and point clouds, making them suitable for real-world applications.

The generated assets can be directly exported to leading game development platforms like Unity and Unreal Engine. This ensures seamless integration into existing production workflows, enabling faster prototyping and iteration for developers.

Tencent has also introduced an interactive “character mode,” allowing users to navigate generated environments in real time with physics and collision features. This adds a layer of realism, making the experience closer to a finished game environment rather than a static prototype.

Expanding use cases and open access strategy

Beyond gaming, HY-World 2.0 supports the creation of digital replicas of real-world environments. By processing videos or images, the model can reconstruct detailed 3D spaces, opening opportunities in sectors such as urban planning, architecture, and cultural preservation.

The company has also highlighted advancements in its underlying technology, including improved scene reconstruction and intelligent navigation within generated environments. These enhancements enable more accurate and complete world-building compared to earlier systems.

By releasing HY-World 2.0 as an open-source platform, Tencent aims to encourage adoption among developers and researchers globally. The move aligns with its broader strategy to expand its AI ecosystem and strengthen its position in next-generation content creation technologies.

As competition intensifies in generative AI for immersive environments, Tencent’s latest offering signals a growing push toward tools that bridge the gap between creative ideas and interactive digital experiences.