Research map / 3D vision / Gaussian splatting
Fast and Compact 3D Gaussian Splatting with Polarized Opacity Prior
Zi-Ming Wang, Kevin Duan, Ko Wei Huang, Akihiro Sugimoto, Shang-Hong Lai
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 Polarized Opacity Prior (POP) instead of opacity reset (ablation) | gsplat-based 3DGS with L2 and blur densification · Mip-NeRF 360 | number of Gaussians: 548k -> 268k | Improved |
| add POP instead of opacity reset (ablation) | gsplat-based 3DGS · Tanks & Temples | PSNR: 22.91 -> 23.62 | Improved |
| add POP instead of opacity reset (SSIM and LPIPS) (ablation) | gsplat-based 3DGS · Mip-NeRF 360 | LPIPS: 0.260 -> 0.277 | Hurt |
| add POP on top of L2 + blur densification (ablation) | gsplat-based 3DGS · Mip-NeRF 360 | number of Gaussians: 583k -> 268k; PSNR 27.22 -> 27.01 | Mixed |
| add blur-based densification to L2 loss (ablation) | gsplat-based 3DGS · Mip-NeRF 360 | PSNR: 26.57 -> 27.22 | Improved |
| replace L1 with L2 loss (ablation) | 3DGS training · Mip-NeRF 360 | PSNR at comparable Gaussian count | Improved |
| extend blur densification to 3k-15k (ablation) | Ours · Mip-NeRF 360, Tanks & Temples, Deep Blending | PSNR: +0.16 on Mip-NeRF 360, -0.07 on T&T, -0.37 on DB | Mixed |
| halt blur densification early (3k-4k) (ablation) | Ours · Mip-NeRF 360 | PSNR: 27.01 -> 26.54 | Hurt |
| compact training vs original 3DGS | Ours · Mip-NeRF 360 | training time (s): 1,304 -> 279 | Improved |
| compact training vs original 3DGS | Ours · Deep Blending | PSNR: 29.847 -> 28.626 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- Mini-Splatting — Mini-Splatting: Representing Scenes with a Constrained Number of Gaussians (ECCV 2024) · better on 1 of 3 shared results
- PUP-3DGS — PUP 3D-GS: Principled Uncertainty Pruning for 3D Gaussian Splatting (CVPR 2025) · better on 2 of 2 shared results
- Speedy-Splat — Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025) · 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
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.94)
- Optimized Minimal 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.93)
- 3D Gaussian Rendering Can Be Sparser: Efficient Rendering via Learned Fragment Pruning (NeurIPS 2024 · similarity 0.93)
- LP-3DGS: Learning to Prune 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.93)
- PUP 3D-GS: Principled Uncertainty Pruning for 3D Gaussian Splatting (CVPR 2025 · similarity 0.93)
- Gaussian Herding across Pens: An Optimal Transport Perspective on Global Gaussian Reduction for 3DGS (NeurIPS 2025 · similarity 0.92)
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.
