Research map / 3D vision / 3D generation
DiffPano++: Scalable and Consistent Multi-View Panorama Generation with Spherical Epipolar-Aware Diffusion
Chenhao Ji, Weicai Ye, Zheng Chen, Junyao Gao, Xiaoshui Huang, Xuekuan Wang, Guofeng Zhang, Songhai Zhang, Tong He, Wanli Ouyang, Cairong Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add spherical epipolar-aware attention instead of MVDream-style 3D attention | DiffPano · multi-view panorama user study | multi-view consistency score: 3.8 -> 4.3 | Improved |
| increase reference views from K=1 to K=3 (S=10) (ablation) | DiffPano · multi-view panorama | PSNR: 29.99 -> 32.29 | Improved |
| increase reference views from K=1 to K=3 (S=10) (ablation) | DiffPano · multi-view panorama | SSIM: 0.69 -> 0.81 | Improved |
| increase reference views from K=3 to K=4 (S=10) (ablation) | DiffPano · multi-view panorama | PSNR: 32.29 -> 33.39 | Mixed |
| increase sample points from S=10 to S=12 (K=4) (ablation) | DiffPano · multi-view panorama | PSNR: 33.39 -> 32.89 | Hurt |
| increase sample points from S=10 to S=12 (K=4) (ablation) | DiffPano · multi-view panorama | FID: 65.98 -> 62.79 | Improved |
| reduce sample points from S=10 to S=6 (K=4) (ablation) | DiffPano · multi-view panorama | inference time (s): 37.91 -> 26.61 | Improved |
| add two-stage training instead of one-stage (ablation) | DiffPano · multi-view panorama | FID: 82.92 -> 74.08 | Hurt |
| add two-stage training instead of one-stage (ablation) | DiffPano · multi-view panorama | IS: 4.52 -> 3.13 | Hurt |
| add two-stage training instead of one-stage (ablation) | DiffPano · multi-view panorama | PSNR: 31.64 -> 31.54 | No clear effect |
| use Pano-SD instead of PanFusion | single-view panorama · PanFusion dataset | inference time (s): 27.6 -> 5.1 | Improved |
| use Pano-SD instead of PanFusion | single-view panorama · PanFusion dataset | FID: 47.62 -> 48.52 | Hurt |
Closest papers by meaning
- DiffPano: Scalable and Consistent Text to Panorama Generation with Spherical Epipolar-Aware Diffusion (NeurIPS 2024 · similarity 0.99)
- CubeDiff: Repurposing Diffusion-Based Image Models for Panorama Generation (ICLR 2025 · similarity 0.90)
- ViewPoint: Panoramic Video Generation with Pretrained Diffusion Models (NeurIPS 2025 · similarity 0.89)
- Taming Stable Diffusion for Text to 360 Panorama Image Generation (CVPR 2024 · similarity 0.89)
- PanoWan: Lifting Diffusion Video Generation Models to 360$^\circ$ with Latitude/Longitude-aware Mechanisms (NeurIPS 2025 · similarity 0.89)
- Conditional Panoramic Image Generation via Masked Autoregressive Modeling (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
