Research map / 3D vision / Gaussian splatting
DR-GS: Physically-Based Deformable and Relightable 2D Gaussians
Jiaxin LI, Tong Wu, Yi Wei, Tailin Wu, Li Zhang
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 learnable normal offset d to Gaussian attributes (ablation) | DR-GS · mesh-driven rendering with extracted mesh | visual quality | Improved |
| add learnable opacity alpha to Gaussian attributes (ablation) | DR-GS · mesh-driven rendering with extracted mesh | visual quality | Improved |
| add denoiser, MIS and mesh acceleration (ablation) | DR-GS · Armadillo | rendering time and PSNR | Improved |
| use particle-driven instead of mesh-driven variant | DR-GS · GlossySynthetic angel (static reconstruction) | PSNR: 25.04 -> 29.64 | Improved |
| use particle-driven instead of mesh-driven variant | DR-GS · GlossySynthetic horse (static reconstruction) | PSNR: 21.25 -> 25.54 | Improved |
| use DR-GS instead of GSP on angel | GSP · GlossySynthetic angel (static reconstruction) | PSNR: 25.70 -> 29.64 | Improved |
| use DR-GS instead of GSP on horse | GSP · GlossySynthetic horse (static reconstruction) | PSNR: 25.90 -> 25.54 | Hurt |
| use DR-GS instead of GS-ROR2 for relighting on angel | GS-ROR2 · GlossySynthetic angel (relighting) | PSNR: 20.81 -> 24.15 | Improved |
| use DR-GS instead of GS-ROR2 for relighting on tbell | GS-ROR2 · GlossySynthetic tbell (relighting) | PSNR: 22.80 -> 22.12 | Hurt |
| use DR-GS instead of GSP after relighting | GSP · User study (deformation) | Physical plausibility: 3.82 -> 4.62 | Improved |
| use DR-GS instead of PhysGaussian | PhysGaussian · User study (deformation) | Physical plausibility: 3.26 -> 4.56 | Improved |
Later papers that compared with this method
2 later accepted papers report DRGS as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Unlocking Accurate Geometry in 3DGS via Spatially Varying Affine Rectification (NeurIPS 2026 · better on 5 of 5 shared results)
- HeroGS: Hierarchical Guidance for Robust 3D Gaussian Splatting under Sparse Views (CVPR 2026 · 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
- Ref-GS: Directional Factorization for 2D Gaussian Splatting (CVPR 2025 · similarity 0.93)
- Reflective Gaussian Splatting (ICLR 2025 · similarity 0.91)
- GI-GS: Global Illumination Decomposition on Gaussian Splatting for Inverse Rendering (ICLR 2025 · similarity 0.91)
- IBGS: Image-Based Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.90)
- GS-IR: 3D Gaussian Splatting for Inverse Rendering (CVPR 2024 · similarity 0.90)
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.
