Research map / 3D vision / Gaussian splatting
StereoGS: Sparse-View 3D Gaussian Splatting via Stereo Priors
Wenhao Yuan, Yiyuan Ge, Deli Cai
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 consistency-aware dense initialization (MVSAnywhere) to vanilla 3DGS (ablation) | 3DGS · LLFF 3-view | PSNR: 16.02 -> 19.75 | Improved |
| add dense initialization to vanilla 3DGS (ablation) | 3DGS · DTU 3-view | PSNR: 10.99 -> 14.10 | Improved |
| add stereo depth regularization on top of dense init (ablation) | StereoGS · LLFF 3-view | PSNR: 19.75 -> 19.79 | No clear effect |
| add stereo depth regularization on top of dense init + opacity decay (ablation) | StereoGS · LLFF 3-view | PSNR: 21.18 -> 21.91 | Improved |
| add gradient-aware opacity decay on top of dense init (ablation) | StereoGS · LLFF 3-view | PSNR: 19.79 -> 21.91 | Improved |
| replace SfM init with MVSAnywhere init (ablation) | StereoGS · LLFF 3-view | PSNR: 18.96 -> 21.91 | Improved |
| replace FoundationStereo with other stereo models (ablation) | StereoGS · LLFF 3-view | PSNR: 21.91 -> 21.28 (LiteAnyStereo), 21.53 (S2M2) | Hurt |
| remove validity mask in stereo depth loss (ablation) | StereoGS · LLFF 3-view | PSNR: 21.91 -> 21.30 | Hurt |
| replace constant decay with relative-gradient exponential decay (ablation) | StereoGS · LLFF 3-view | PSNR: 21.45 -> 21.91 | Improved |
| use step or linear function of relative gradient instead of constant decay (ablation) | StereoGS · LLFF 3-view | PSNR: 21.45 -> 21.17 / 21.24 | Hurt |
| use raw gradient instead of relative gradient in exponential decay (ablation) | StereoGS · LLFF 3-view | PSNR: 21.91 -> 21.41 | Hurt |
| use StereoGS instead of Binocular3DGS | StereoGS · LLFF 3-view | PSNR: 21.44 -> 21.91 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DropGaussian — DropGaussian: Structural Regularization for Sparse-view Gaussian Splatting (CVPR 2025) · better on 2 of 2 shared results
- D2GS — D$^2$GS: Depth-and-Density Guided Gaussian Splatting for Stable and Accurate Sparse-View Reconstruction (ICLR 2026) · better on 1 of 1 shared result
- CoR-GS — CoR-GS: Sparse-View 3D Gaussian Splatting via Co-Regularization (ECCV 2024) · 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
- NexusGS: Sparse View Synthesis with Epipolar Depth Priors in 3D Gaussian Splatting (CVPR 2025 · similarity 0.92)
- Binocular-Guided 3D Gaussian Splatting with View Consistency for Sparse View Synthesis (NeurIPS 2024 · similarity 0.92)
- Quantifying and Alleviating Co-Adaptation in Sparse-View 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views (CVPR 2025 · similarity 0.91)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.91)
- GaussianPro: 3D Gaussian Splatting with Progressive Propagation (ICML 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.
