Research map / 3D vision / Gaussian splatting
Revisiting Photometric Ambiguity for Accurate Gaussian-Splatting Surface Reconstruction
Jiahe Li, Jiawei Zhang, Xiao Bai, Jin Zheng, Xiaohan Yu, Lin Gu, Gim Hee Lee
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 Gaussian primitive truncation (ablation) | AmbiSuR (PGSR base with depth prior) · Tanks and Temples | F1-score: 0.547 -> 0.558 | Improved |
| add ray-color consistency (ablation) | AmbiSuR · Tanks and Temples | F1-score: 0.558 -> 0.566 | Improved |
| add SH ambiguity indication instead of naive uniform depth-normal loss (ablation) | AmbiSuR · Tanks and Temples | F1-score: 0.557 -> 0.589 | Improved |
| use naive uniform depth-normal loss instead of SH indication (ablation) | AmbiSuR with truncation and ray-color · Tanks and Temples | Precision: 0.552 -> 0.533 | Hurt |
| add dual-end (upper and lower) indication (ablation) | SH ambiguity indicator · DTU | Chamfer distance: 0.473 -> 0.461 | Improved |
| add amorphous mask (ablation) | SH ambiguity indicator · DTU | Chamfer distance: 0.472 -> 0.461 | Improved |
| add parameter separation (ablation) | SH ambiguity indicator · DTU | Chamfer distance: 0.470 -> 0.461 | Improved |
| truncation bound gamma 1.5 sigma instead of 2.0 sigma (ablation) | Gaussian primitive truncation · Tanks and Temples | F1-score: 0.589 -> 0.574 | Hurt |
| no truncation (gamma infinite) versus gamma 2.0 (ablation) | Gaussian primitive truncation · Tanks and Temples | F1-score: 0.576 -> 0.589 | Improved |
| vary upper/lower indicator sizes (90-98% and 5-15%) (ablation) | Dual-end indication · DTU | Chamfer distance: 0.461 -> 0.463 | No clear effect |
| swap depth prior model (VGGT, MVSAnywhere, DepthPro, DA3, DAV2) (ablation) | AmbiSuR · DTU | Chamfer distance: 0.468 (VGGT) vs 0.461 (DA3) | No clear effect |
| use AmbiSuR instead of PGSR with same priors (ablation) | PGSR baseline · DTU | Chamfer distance: 0.480 -> 0.461 | Improved |
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.
- GeoSVR — GeoSVR: Taming Sparse Voxels for Geometrically Accurate Surface Reconstruction (NeurIPS 2025) · better on 2 of 2 shared results
- Geo-NeuS — Geo-Neus: Geometry-Consistent Neural Implicit Surfaces Learning for Multi-view Reconstruction (NeurIPS 2022) · 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
- VA-GS: Enhancing the Geometric Representation of Gaussian Splatting via View Alignment (NeurIPS 2025 · similarity 0.91)
- FatesGS: Fast and Accurate Sparse-View Surface Reconstruction Using Gaussian Splatting with Depth-Feature Consistency (AAAI 2025 · similarity 0.89)
- Subsurface Scattering for Gaussian Splatting (NeurIPS 2024 · similarity 0.89)
- 3D Convex Splatting: Radiance Field Rendering with 3D Smooth Convexes (CVPR 2025 · similarity 0.88)
- Evolving High-Quality Rendering and Reconstruction in a Unified Framework with Contribution-Adaptive Regularization (CVPR 2025 · similarity 0.88)
- Ref-GS: Directional Factorization for 2D Gaussian Splatting (CVPR 2025 · similarity 0.88)
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.
