Research map / 3D vision / Gaussian splatting
GeoGaussian: Geometry-aware Gaussian Splatting for Scene Rendering
Yanyan Li, Chenyu Lyu, Yan Di, Guangyao Zhai, Gim Hee Lee, Federico Tombari
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 thin surface-aligned Gaussians with tangent-space densification and co-planar smoothness loss | 3DGS · Replica (avg, 10% train) | PSNR: 30.62 -> 31.83 | Improved |
| use geometry-aware Gaussians at full training data | 3DGS · Replica (avg, 100% train) | PSNR: 37.84 -> 38.20 | Improved |
| use geometry-aware Gaussians on a low-texture TUM sequence with sparse views | 3DGS · TUM f3/cabinet (10%) | PSNR: 22.43 -> 25.59 | Improved |
| use geometry-aware Gaussians on TUM RGB-D | 3DGS · TUM RGB-D (avg) | PSNR: 24.73 -> 26.31 | Improved |
| geometry-aware method on lag-cabinet sequences | 3DGS and LightGS · TUM lag-cabinet | rendering quality | Hurt |
| reduce training views from 50% to 10% (ablation) | 3DGS · Replica R1 | PSNR: 37.45 -> 30.49 | Hurt |
| reduce training views from 50% to 10% (ablation) | GeoGaussian · Replica R1 | PSNR: 38.00 -> 31.65 | Hurt |
| geometry-aware constraints with full training views on OFF3 | 3DGS · Replica OFF3 (100%) | PSNR: 36.56 -> 36.66 | No clear effect |
| geometry-aware Gaussians in place of random densification | 3DGS · Replica (number of Gaussians) | number of Gaussians | Improved |
Later papers that compared with this method
1 later accepted paper reports GeoGaussian as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- EnerGS: Energy-Based Gaussian Splatting under Partial Geometric Priors (ICML 2026 · better on 0 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
- CityGaussianV2: Efficient and Geometrically Accurate Reconstruction for Large-Scale Scenes (ICLR 2025 · similarity 0.92)
- LeanGaussian: Breaking Pixel or Point Cloud Correspondence in Modeling 3D Gaussians (CVPR 2025 · similarity 0.92)
- VastGaussian: Vast 3D Gaussians for Large Scene Reconstruction (CVPR 2024 · similarity 0.91)
- MaskGaussian: Adaptive 3D Gaussian Representation from Probabilistic Masks (CVPR 2025 · similarity 0.91)
- DropGaussian: Structural Regularization for Sparse-view Gaussian Splatting (CVPR 2025 · similarity 0.91)
- DreamGaussian: Generative Gaussian Splatting for Efficient 3D Content Creation (ICLR 2024 · similarity 0.91)
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.
