Research map / 3D vision / Gaussian splatting
MVSGaussian: Fast Generalizable Gaussian Splatting Reconstruction from Multi-View Stereo
Tianqi Liu, Guangcong Wang, Shoukang Hu, Liao Shen, Xinyi Ye, Yuhang Zang, Zhiguo Cao, Wei Li, Ziwei Liu
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 cascade structure (ablation) | MVSGaussian · DTU | PSNR: 26.71 -> 27.48 | Improved |
| add depth-aware volume rendering to splatting (hybrid rendering) (ablation) | MVSGaussian · DTU | PSNR: 27.48 -> 28.21 | Improved |
| add depth-aware volume rendering to splatting (hybrid rendering) (ablation) | MVSGaussian · Tanks and Temples | PSNR: 21.70 -> 23.28 | Improved |
| decode RGB directly instead of SH coefficients (ablation) | MVSGaussian · NeRF Synthetic | PSNR: 24.27 -> 26.46 | Improved |
| decode RGB directly instead of SH coefficients (ablation) | MVSGaussian · DTU | PSNR: 28.19 -> 28.21 | No clear effect |
| use multi-view consistency check instead of direct concatenation for initialization (ablation) | per-scene optimization · Real Forward-facing | PSNR: 26.18 -> 26.98 | Improved |
| use consistency check instead of voxel downsampling (ablation) | per-scene optimization · Real Forward-facing | PSNR: 26.72 -> 26.98 | Improved |
| add density prediction for volume rendering (ablation) | MVSGaussian · NeRF Synthetic | PSNR: 26.22 -> 26.46 | Improved |
| increase input views from 2 to 4 (ablation) | MVSGaussian · DTU | PSNR: 25.78 -> 28.43 | Improved |
| sample 2 Gaussians per pixel instead of 1 (ablation) | MVSGaussian · DTU | PSNR: 28.21 -> 28.26 (memory 0.876 -> 1.508 GB) | No clear effect |
| initialize 3D-GS from generalizable model instead of COLMAP | 3D-GS · Real Forward-facing | PSNR / fine-tuning time: 23.92 at 10min -> 26.98 at 45s | Improved |
| use MVS-derived pixel-aligned Gaussians instead of epipolar Transformer (PixelSplat) | generalizable Gaussian Splatting · NeRF Synthetic 2-view | PSNR: 15.77 -> 25.34 | 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.
- PixelSplat — pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024) · better on 2 of 2 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
- PUP 3D-GS: Principled Uncertainty Pruning for 3D Gaussian Splatting (CVPR 2025 · similarity 0.91)
- GaussianPro: 3D Gaussian Splatting with Progressive Propagation (ICML 2024 · similarity 0.91)
- FreeSplat: Generalizable 3D Gaussian Splatting Towards Free View Synthesis of Indoor Scenes (NeurIPS 2024 · similarity 0.91)
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.91)
- Evolving High-Quality Rendering and Reconstruction in a Unified Framework with Contribution-Adaptive Regularization (CVPR 2025 · similarity 0.91)
- DOGS: Distributed-Oriented Gaussian Splatting for Large-Scale 3D Reconstruction Via Gaussian Consensus (NeurIPS 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.
