Research map / 3D vision / Gaussian splatting
Tackling Misattribution in 3D Intrinsic Decomposition via Proximity Attention Point Rendering
Alireza Moazeni, Shichong Peng, Yanshu Zhang, Chirag Vashist, Ke Li
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 albedo/shading decomposition with deterministic 2D prior (ablation) | PAPR · synthetic scenes avg | NVS PSNR: 32.84 -> 33.71 | Improved |
| Adapt prior with cIMLE (M=10) (ablation) | Intrinsic PAPR · synthetic scenes avg | NVS PSNR: 33.71 -> 34.36 | Improved |
| Add space-carving loss (ablation) | Intrinsic PAPR · synthetic scenes avg | NVS PSNR: 34.36 -> 34.83 | Improved |
| Add per-view scalar (ablation) | Intrinsic PAPR · synthetic scenes avg | NVS PSNR: 34.83 -> 35.00 | Improved |
| Add space-carving loss (ablation) | Intrinsic PAPR · real scenes (T&T + Mip-NeRF 360) | PSNR: 29.14 -> 29.99 | Improved |
| Add space-carving loss (ablation) | Intrinsic PAPR · real scenes (T&T + Mip-NeRF 360) | LPIPS: 0.126 -> 0.094 | Improved |
| Add cIMLE and space-carving filter chain (ablation) | Intrinsic PAPR · synthetic scenes avg | albedo PSNR: 29.11 -> 31.21 | Improved |
| Add cIMLE and space-carving filter chain (ablation) | Intrinsic PAPR · synthetic scenes avg | MACE: 0.019 -> 0.015 | Improved |
| Give baselines the same pseudo-GT albedo supervision (ablation) | GS-IR · 4 scenes avg NVS | PSNR: 29.66 -> 27.05 | Hurt |
| Give baselines the same pseudo-GT albedo supervision (ablation) | IRGS · 4 scenes avg NVS | PSNR: 30.06 -> 26.80 | Hurt |
| Give baselines the same pseudo-GT albedo supervision (ablation) | Intrinsic-NeRF · 4 scenes avg NVS | PSNR: 29.18 -> 28.97 | No clear effect |
| Use more training views (25 to 200) (ablation) | Intrinsic PAPR vs GS-IR · albedo transfer MSE | ratio GS-IR/ours: 5.10x (25 views) -> 5.23x (200 views) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GS-IR — GS-IR: 3D Gaussian Splatting for Inverse Rendering (CVPR 2024) · better on 3 of 3 shared results
- IRGS — IRGS: Inter-Reflective Gaussian Splatting with 2D Gaussian Ray Tracing (CVPR 2025) · better on 2 of 2 shared results
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
- PAPR: Proximity Attention Point Rendering (NeurIPS 2023 · similarity 0.90)
- Latent Intrinsics Emerge from Training to Relight (NeurIPS 2024 · similarity 0.87)
- PAPR in Motion: Seamless Point-level 3D Scene Interpolation (CVPR 2024 · similarity 0.87)
- IDArb: Intrinsic Decomposition for Arbitrary Number of Input Views and Illuminations (ICLR 2025 · similarity 0.87)
- Differentiable Point-based Inverse Rendering (CVPR 2024 · similarity 0.86)
- CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction (NeurIPS 2026 · 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.
