HY-World 2.0 enables AI to generate navigable 3D scenes from text, images, or videos using advanced multi-modal processing

Accepts diverse inputs including text prompts, single-view images, multi-view images, and videos to create 3D world representations

Generates high-fidelity 3D Gaussian Splatting (3DGS) scenes through a four-stage pipeline: panorama generation, trajectory planning, world expansion, and composition

Introduces upgraded components including HY-Pano 2.0 for improved panorama fidelity, WorldStereo 2.0 with consistent memory for keyframe-based view generation, and WorldMirror 2.0 for universal 3D prediction

Represents an advancement over the original HY-World 1.0 with enhanced 3D scene understanding and planning capabilities