Tencent Open Sources Huan Yuan 3D World Model 2.0 for One-Click Editable 3D Space Assets
Tencent's Huan Yuan team today announced the release and open-sourcing of "Huan Yuan 3D World Model 2.0" (HY-World2.0), a major upgrade to its multimodal world model. This version marks a shift from simple video generation to creating interactive, editable 3D assets. It supports text, image, and video inputs, enabling the generation, reconstruction, and simulation of complete 3D spaces that include people, objects, and scenes. The model also exports in multiple formats such as Mesh, 3DGS, and point clouds, and integrates seamlessly with mainstream game engines like Unity and Unreal Engine.

On the technical side, HY-World2.0 unifies spatial understanding, generation, and reconstruction into a single process. Key upgrades include the HY-Pano-2.0 model, which uses an end-to-end implicit learning approach to generate 360-degree panoramic maps without requiring camera parameters; a self-developed spatial agent that combines VLM with navmesh representation, enabling intelligent roaming trajectory planning; and the WorldStereo mechanism, which ensures high consistency in geometry and visual appearance between newly generated areas and existing scenes.


Additionally, the upgraded WorldMirror2.0 architecture replicates real-world scenes by predicting dense point clouds and camera parameters in a single pass, enabling high-precision digital twin construction.
Compared to mainstream models like Google's Genie3, Huan Yuan 2.0 stands out because its generated assets feature realistic physical collision properties and support free exploration in character mode. This significantly lowers the barrier to building game level prototypes and embodied intelligent simulation environments.
This release marks an evolution for Tencent Huan Yuan from "manipulating objects" to "creating worlds" in 3D generation. With a state-of-the-art spatial consistency memory mechanism, it pushes AI-driven 3D content creation into a practical, real-world phase.
Related article
Meta Removes AI Photo Editing Feature Following User Backlash
Meta, the social media giant, is once again embroiled in a public debate regarding the delicate balance between artificial intelligence and user privacy. According to TechCrunch, Meta’s Superintelligence Labs introduced a new AI image generator, Muse
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage
Related Special Topic Recommendations
Comments (0)
0/500
Tencent's Huan Yuan team today announced the release and open-sourcing of "Huan Yuan 3D World Model 2.0" (HY-World2.0), a major upgrade to its multimodal world model. This version marks a shift from simple video generation to creating interactive, editable 3D assets. It supports text, image, and video inputs, enabling the generation, reconstruction, and simulation of complete 3D spaces that include people, objects, and scenes. The model also exports in multiple formats such as Mesh, 3DGS, and point clouds, and integrates seamlessly with mainstream game engines like Unity and Unreal Engine.

On the technical side, HY-World2.0 unifies spatial understanding, generation, and reconstruction into a single process. Key upgrades include the HY-Pano-2.0 model, which uses an end-to-end implicit learning approach to generate 360-degree panoramic maps without requiring camera parameters; a self-developed spatial agent that combines VLM with navmesh representation, enabling intelligent roaming trajectory planning; and the WorldStereo mechanism, which ensures high consistency in geometry and visual appearance between newly generated areas and existing scenes.


Additionally, the upgraded WorldMirror2.0 architecture replicates real-world scenes by predicting dense point clouds and camera parameters in a single pass, enabling high-precision digital twin construction.
Compared to mainstream models like Google's Genie3, Huan Yuan 2.0 stands out because its generated assets feature realistic physical collision properties and support free exploration in character mode. This significantly lowers the barrier to building game level prototypes and embodied intelligent simulation environments.
This release marks an evolution for Tencent Huan Yuan from "manipulating objects" to "creating worlds" in 3D generation. With a state-of-the-art spatial consistency memory mechanism, it pushes AI-driven 3D content creation into a practical, real-world phase.
Meta Removes AI Photo Editing Feature Following User Backlash
Meta, the social media giant, is once again embroiled in a public debate regarding the delicate balance between artificial intelligence and user privacy. According to TechCrunch, Meta’s Superintelligence Labs introduced a new AI image generator, Muse
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage





Home






