arxivcs.CVcs.AI2026-07-01
Pano2World: End-to-End 3D Generation via Unified Multi-View Sequences
Zhenjia Li, Jinrang Jia, Yifeng Shi
A single panorama captures the full visual sphere from one camera center, yet confines users to looking around in place without enabling true scene exploration. Converting a single panorama into a persistent, renderable 3D representation for free-viewpoint navigation has attracte…