Research map / 3D vision / 3D generation
Structural Multiplane Image: Bridging Neural View Synthesis and 3D Reconstruction
Mingfang Zhang, Jinglu Wang, Xiao Li, Yifei Huang, Yoichi Sato, Yan Lu
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 |
|---|---|---|---|
| Use scene-adaptive structural planes instead of fixed fronto-parallel MPI layers | S-MPI · ScanNet, n<30 | PSNR: 21.058 -> 23.604 | Improved |
| Use global proxy embeddings to enforce multi-view consistency (ablation) | S-MPI, 100 queries · ScanNet, multi-view view synthesis ablation | PSNR: 22.65 -> 23.11 | Improved |
| Increase query number from 25 to 100 while retaining MVC (ablation) | S-MPI · ScanNet, multi-view view synthesis ablation | PSNR: 22.97 -> 23.11 | Improved |
| Increase the input-view gap from 0 to 20 frames (ablation) | S-MPI · ScanNet, planar recall | RC^0.1m: 52.61 -> 52.95 | Improved |
| Increase the input-view gap from 20 to 40 frames (ablation) | S-MPI · ScanNet, planar recall | RC^0.1m: 52.95 -> 46.43 | Hurt |
| Increase planar area in the scene (ablation) | S-MPI · ScanNet, view synthesis | PSNR: 22.97 -> 24.52 | Improved |
| Set planar proxy count to zero, treating all regions as non-planar (ablation) | S-MPI · ScanNet, view synthesis | PSNR: 24.52 -> 21.10 | Hurt |
| Use S-MPI instead of PlaneMVS | Multi-view reconstruction · ScanNet | rmse: 0.186 -> 0.205 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- PlaneMVS — PlaneMVS: 3D Plane Reconstruction From Multi-View Stereo (CVPR 2022) · better on 1 of 2 shared results
- DP-NeRF (2) — DP-NeRF: Deblurred Neural Radiance Field With Physical Scene Priors (CVPR 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Tiled Multiplane Images for Practical 3D Photography (ICCV 2023 · similarity 0.90)
- Pushing the Boundaries of View Extrapolation With Multiplane Images (CVPR 2019 · similarity 0.89)
- GTR: Improving Large 3D Reconstruction Models through Geometry and Texture Refinement (ICLR 2025 · similarity 0.86)
- Towards In-the-wild 3D Plane Reconstruction from a Single Image (CVPR 2025 · similarity 0.86)
- PMNI: Pose-free Multi-view Normal Integration for Reflective and Textureless Surface Reconstruction (CVPR 2025 · similarity 0.85)
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 2024 · similarity 0.85)
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.
