3D-aware 360° video diffusion model that generates panoramic videos along user-defined camera paths from sparse panoramas.
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.
Pantheon360 is a research video diffusion model for generating digital twins from sparse 360° inputs. It was developed by researchers at the University of Southern California, National Yang Ming Chiao Tung University, Cornell University and Bosch Research, and presented at CVPR 2026.
Given a few panoramas, the system crops them into perspective views and reconstructs a 3D point cloud "cache" with foundation models such as PI3 and VGGT. The cache is rendered along a user-defined camera trajectory as a geometry-only equirectangular video. This video conditions a fine-tuned video diffusion model, together with CLIP features from perspective crops of the first frame. The diffusion model then refines texture while the cache enforces geometric consistency. For interpolation between two panoramas, it uses dual-anchor latent fusion.
The project page lists use cases including two-view 360° interpolation, chaining multiple panoramas for extended trajectories, single-view camera exploration, and stabilizing shaky 360° footage through trajectory replanning. The authors compare it against the perspective-based GEN3C. Code is listed as "coming soon"; no public weights, demo or license were found, so it is currently a research release only.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
360-degree video generation model for digital twins that takes several panoramas plus a camera path and uses a reconstructed 3D point cloud to keep the panoramic video consistent. Code is coming soon.
Closest neighbours by meaning, not just by category.
Research
Multi-view video diffusion model that turns monocular or sparse videos of dynamic humans into dense multi-view videos for 4D Gaussian splats.
Model · Meta
Meta and HKUST research method for fast artist-like 3D mesh generation using a MeshVAE and flow-based diffusion transformer.
Research
Diffusion framework that learns text-to-3D human motion generation from 2D poses extracted from monocular videos.
Research
Research system that turns a single image into a 360° Gaussian scaffold and renders interactive video at 8 FPS.
Model
Research model for video relighting that jointly generates relit video and per-frame environment maps, no camera poses needed.
Model
Research generative world model that creates consistent whole-house panorama tours from a floorplan and style reference.