Research map / 3D vision / NeRF and radiance fields
Deep Multi Depth Panoramas for View Synthesis
Kai-En Lin, Zexiang Xu, Ben Mildenhall, Pratul P. Srinivasan, Yannick Hold-Geoffroy, Stephen DiVerdi, Qi Sun, Kalyan Sunkavalli, Ravi Ramamoorthi
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 |
|---|---|---|---|
| increase MDP layers from 1 to 5 (ablation) | MDP · Synthetic scenes (MDP layer count) | PSNR: 25.75 -> 26.39 | Improved |
| increase MDP layers from 1 to 5 (ablation) | MDP · Synthetic scenes (MDP layer count) | SSIM: 0.8565 -> 0.8664 | Improved |
| increase MDP layers from 1 to 5 (ablation) | MDP · Synthetic scenes (MDP layer count) | L1: 0.0269 -> 0.0251 | Improved |
| increase MDP layers from 4 to 5 (ablation) | MDP · Synthetic scenes (MDP layer count) | PSNR: 26.35 -> 26.39 | Improved |
| increase MDP layers from 2 to 5 (ablation) | MDP · Synthetic scenes (MDP layer count) | PSNR: 26.17 -> 26.39 | Improved |
| use learned MDP instead of per-view MPI blending | MPI · Synthetic scenes (MPI comparison) | PSNR: 24.75 -> 26.39 | Improved |
| use learned MDP instead of per-view MPI blending | MPI · Synthetic scenes (MPI comparison) | Storage: 3.355GB -> 0.165GB | Improved |
| use single layer MDP depth instead of prior depth estimator, as points | Depth as points · Synthetic scenes (single RGBD panorama 6DoF) | PSNR: 23.06 -> 25.75 | Improved |
| use own depth in place of prior depth with disocclusion method | Depth + disocclusion method · Synthetic scenes (single RGBD panorama 6DoF) | PSNR: 23.20 -> 24.98 | Improved |
| use 5-layer MDPs instead of baseline at max disparity 128 | Depth + disocclusion method · Synthetic scenes (max disparity 128) | PSNR: 18.17 -> 21.94 | Improved |
Closest papers by meaning
- MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024 · similarity 0.88)
- Zero-Shot Novel View and Depth Synthesis with Multi-View Geometric Diffusion (CVPR 2025 · similarity 0.87)
- CompNVS: Novel View Synthesis with Scene Completion (ECCV 2022 · similarity 0.87)
- Snap Angle Prediction for 360° Panoramas (ECCV 2018 · similarity 0.86)
- BIPS: Bi-modal Indoor Panorama Synthesis via Residual Depth-Aided Adversarial Learning (ECCV 2022 · similarity 0.86)
- MVD-Fusion: Single-view 3D via Depth-consistent Multi-view Generation (CVPR 2024 · similarity 0.86)
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.
