Research map / 3D vision / Gaussian splatting
FSFSplatter: Geometrically Accurate Reconstruction with Free Sparse-view Images within 2 minutes
Yibin Zhao, Yihan Pan, Jun Nan, Liwei Chen, Jianjun Yi
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 monocular depth ranking loss Lrank (ablation) | FSFSplatter optimization · DTU | Chamfer distance (mm): 2.905 -> 1.581 | Improved |
| add depth smoothness loss Lsmooth (ablation) | FSFSplatter optimization · DTU | Chamfer distance (mm): 2.653 -> 1.581 | Improved |
| add multi-view feature loss LMVS (ablation) | FSFSplatter optimization · DTU | Chamfer distance (mm): 3.130 -> 1.581 | Improved |
| make camera poses differentiable (ablation) | FSFSplatter optimization · DTU | Chamfer distance (mm): 2.855 -> 1.581 | Improved |
| add contribution-based pruning (ablation) | FSFSplatter initialization · DTU | Chamfer distance (mm): 2.930 -> 1.581 | Improved |
| add self-splitting Gaussian densification (ablation) | FSFSplatter initialization · Replica | Chamfer distance (mm): 59.83 -> 33.66 | Improved |
| add self-splitting Gaussian densification (ablation) | FSFSplatter initialization · DTU | Chamfer distance (mm): 3.584 -> 1.581 | Improved |
| remove extra supervision in per-scene optimization (ablation) | FSFSplatter · DTU | Chamfer distance (mm): 2.208 (wo Opt.) vs 2.905 (No Lrank) | Hurt |
| add per-scene optimization after initialization | FSFSplatter · DTU | Chamfer distance (mm): 1.907 -> 1.581 | Improved |
| add per-scene optimization after initialization | FSFSplatter · Replica | PSNR (dB): 23.51 -> 35.79 | Improved |
| increase iterations of per-scene optimization under sparse views | 3DGS-based baselines · DTU | Chamfer distance | Hurt |
| use FSFSplatter initialization instead of VGGT point cloud + GS init | FSFSplatter · DTU | surface reconstruction: 26.3% improvement; 11.9dB PSNR increase | 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.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 6 of 6 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
- DeSplat: Decomposed Gaussian Splatting for Distractor-Free Rendering (CVPR 2025 · similarity 0.89)
- FatesGS: Fast and Accurate Sparse-View Surface Reconstruction Using Gaussian Splatting with Depth-Feature Consistency (AAAI 2025 · similarity 0.89)
- CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction (NeurIPS 2026 · similarity 0.89)
- FreeSplat: Generalizable 3D Gaussian Splatting Towards Free View Synthesis of Indoor Scenes (NeurIPS 2024 · similarity 0.88)
- Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views (CVPR 2025 · similarity 0.88)
- Generative Sparse-View Gaussian Splatting (CVPR 2025 · similarity 0.88)
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.
