Post
2557
Just tried tencent/HY-World-2.0 ā a multimodal world model that takes in text or a single image and generates editable 3D scenes.
Unlike Google's Genie and HY-World 1.5, v2.0 generates engine-ready 3D content:
š® Direct import into Unreal Engine and Unity ā no format wrangling
š§ Supports multiple 3D asset formats: Mesh, 3DGS, point cloud, etc.
āļø Fully editable ā not a baked video, but actual geometry you can modify
š¤ Also usable for embodied simulation environments
Basically: from "AI generates a world you can look at" ā "AI generates a world you can ship."
Unlike Google's Genie and HY-World 1.5, v2.0 generates engine-ready 3D content:
š® Direct import into Unreal Engine and Unity ā no format wrangling
š§ Supports multiple 3D asset formats: Mesh, 3DGS, point cloud, etc.
āļø Fully editable ā not a baked video, but actual geometry you can modify
š¤ Also usable for embodied simulation environments
Basically: from "AI generates a world you can look at" ā "AI generates a world you can ship."