Research map / 3D vision / Gaussian splatting
GAINS: Gaussian-based Inverse Rendering from Sparse Multi-View Captures
Patrick Noras, Jun Myeong Choi, Didier Stricker, Pieter Peers, Roni Sengupta
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 SDS diffusion guidance in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | NVS PSNR: 23.57 -> 23.67 | Improved |
| add outlier removal in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | NVS PSNR: 21.75 -> 23.57 | Improved |
| add normal guidance in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | NVS PSNR: 19.42 -> 21.75 | Improved |
| add depth guidance in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | NVS PSNR: 16.73 -> 19.42 | Improved |
| add depth guidance in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | Normal MAE: 32.23 -> 19.38 | Improved |
| add normal guidance in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | Normal MAE: 19.38 -> 13.79 | Improved |
| add SDS in Stage I (ablation) | GAINS Stage I · Shiny Blender (8 views) | Relight PSNR: 21.91 -> 21.96 | Improved |
| add IID prior in Stage II (ablation) | GAINS Stage II · Shiny Blender (8 views) | albedo/relighting | Improved |
| add segmentation prior in Stage II (ablation) | GAINS Stage II · Shiny Blender (8 views) | NVS/specular | Improved |
| add multi-illuminated diffusion prior in Stage II (ablation) | GAINS Stage II · Shiny Blender (8 views) | all tasks | Improved |
| use GAINS instead of Ref-Gaussian | relighting · Synthetic4Relight (8 views) | Relight PSNR: 24.29 -> 28.16 | Improved |
| use GAINS instead of Ref-Gaussian | relighting · Shiny Blender (8 views) | Relight PSNR: 17.26 -> 22.29 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GI-GS — GI-GS: Global Illumination Decomposition on Gaussian Splatting for Inverse Rendering (ICLR 2025) · better on 8 of 8 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
- MaterialRefGS: Reflective Gaussian Splatting with Multi-view Consistent Material Inference (NeurIPS 2025 · similarity 0.91)
- CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction (NeurIPS 2026 · similarity 0.91)
- Reflective Gaussian Splatting (ICLR 2025 · similarity 0.91)
- Quantifying and Alleviating Co-Adaptation in Sparse-View 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.90)
- Inverse Rendering using Multi-Bounce Path Tracing and Reservoir Sampling (ICLR 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.
