Research generative world model that creates consistent whole-house panorama tours from a floorplan and style reference.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
PanoWorld is a research model for generating a consistent whole-house VR tour from a floorplan and a style reference. It treats whole-house synthesis as autoregressive generation of node-based 360-degree panoramas, matching the discrete navigation used in real VR tour products.
It uses a floorplan-derived 3D shell as a global geometric proxy and a dynamic 3D Gaussian Splatting cache as renderable spatial memory. A feed-forward panoramic large reconstruction model (LRM) lifts generated panoramas into local 3DGS updates. Room-aware Group Attention limits feature interference between rooms, and a topology-aware progressive caching strategy fuses updates without reconstructing the full history. The authors report improved cross-node layout and material consistency compared with adapted baselines.
The project page offers an interactive panorama viewer and qualitative and quantitative comparisons (HPSv3, CLIP-I Style, overlap PSNR). Code and models had not been released according to creator coverage, and no code or weights links were found on the project page. The paper is on arXiv (2605.17916) by Jinrang Jia, Zhenjia Li, Yijiang Hu and Yifeng Shi.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Generative world model that turns a floor plan and style reference into consistent panoramic tours of a full house, using a 3D shell as memory. Code and models not yet released.
Closest neighbours by meaning, not just by category.
Model
3D-aware 360° video diffusion model that generates panoramic videos along user-defined camera paths from sparse panoramas.
Model · World Labs
World Labs' world model that generates and reconstructs explorable 3D worlds from images, text, video or 3D input.
Model
Research model generating 3D indoor scenes as sparse voxels, refined from 25 cm to 1.5625 cm by an autoregressive cascade.
Research · Tencent
Tencent Hunyuan research framework that turns one text prompt into an explorable, editable open-world 3D scene.
Model
Research model that unifies 3D scene generation and reconstruction in a single pixel-space diffusion model.
Model
Research framework that turns posed images with instance masks and 3D boxes into per-object meshes for cluttered scenes.
