Research map / 3D vision / Gaussian splatting
UniGS: Modeling Unitary 3D Gaussians for Novel View Synthesis from Sparse-view Images
Jiamin Wu, Kenkun Liu, Xiaoke Jiang, Yuan Yao, Lei 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 |
|---|---|---|---|
| model unitary 3D Gaussians in world space with a DETR-like decoder instead of per-pixel local prediction and concatenation (ablation) | UniGS · GSO 4 views | PSNR: 25.6033 -> 26.5334 (Objaverse val, UNet only vs full) | Improved |
| initialize 3D Gaussians randomly with no constraint instead of regressing from image features (ablation) | UniGS · Objaverse validation | PSNR: 26.5334 -> 12.1213 | Hurt |
| random initialization inside the cone of vision (ablation) | UniGS · Objaverse validation | PSNR: 26.5334 -> 22.6740 | Hurt |
| add cross-view attention in the feature extractor (ablation) | UniGS · Objaverse validation | PSNR: 25.3923 -> 26.5334 | Improved |
| add camera modulation on queries in MVDFA (ablation) | UniGS · Objaverse validation | PSNR: 26.1328 -> 26.5334 | Improved |
| sample per-view points instead of shared 3D sampling points (ablation) | UniGS · Objaverse validation | PSNR: 25.8392 -> 26.5334 | Improved |
| increase number of Gaussians beyond 19600 (ablation) | UniGS · Objaverse validation | PSNR | No clear effect |
| raise key/value downsampling rate in SESA self-attention (ablation) | UniGS · Objaverse validation | memory and PSNR | Mixed |
| UniGS vs best prior feed-forward method on GSO 4 views | UniGS · GSO 4 fixed views | PSNR: 26.2487 -> 30.4245 (vs LGM Large); 30.0500 (GRM) | Improved |
| train on 4 views, test on varying number of views | UniGS · GSO | PSNR | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Splatter Image — Splatter Image: Ultra-Fast Single-View 3D Reconstruction (CVPR 2024) · better on 2 of 2 shared results
- DreamGaussian — DreamGaussian: Generative Gaussian Splatting for Efficient 3D Content Creation (ICLR 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
- No Pose, No Problem: Surprisingly Simple 3D Gaussian Splats from Sparse Unposed Images (ICLR 2025 · similarity 0.90)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.90)
- UniGS: Unified Language-Image-3D Pretraining with Gaussian Splatting (ICLR 2025 · similarity 0.90)
- FewViewGS: Gaussian Splatting with Few View Matching and Multi-stage Training (NeurIPS 2024 · similarity 0.89)
- S2Gaussian: Sparse-View Super-Resolution 3D Gaussian Splatting (CVPR 2025 · similarity 0.89)
- FreeSplat: Generalizable 3D Gaussian Splatting Towards Free View Synthesis of Indoor Scenes (NeurIPS 2024 · similarity 0.89)
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.
