AI Paper Atlas

Research map / 3D vision / Gaussian splatting

Mini-Splatting: Representing Scenes with a Constrained Number of Gaussians

Guangchi Fang, Bing Wang

ECCV 2024 · 3D vision › Gaussian splatting · approach: Compact and efficient Gaussian splatting · 224 citations (OpenAlex)

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 changedWhereMeasuredEffect
Add blur split and depth reinitialization (Mini-Splatting-D densification) (ablation)3DGS · Mip-NeRF360LPIPS: 0.216 -> 0.175Improved
Add blur split to baseline 3DGS (ablation)3DGS · Mip-NeRF360PSNR: 27.47 -> 27.47No clear effect
Reinitialize with alpha-blended depth instead of mid-point depth (ablation)Mini-Splatting-D · Mip-NeRF360PSNR: 27.54 -> 17.67Hurt
Use Gaussian center depth instead of mid-point depth (ablation)Mini-Splatting-D · Mip-NeRF360PSNR: 27.54 -> 27.57No clear effect
Add intersection preserving and importance-weighted sampling instead of direct pruning (ablation)Mini-Splatting-D · Mip-NeRF360quality vs number of GaussiansImproved
Prune 3DGS directly from 6.1M to 0.6M Gaussians (ablation)3DGS · Mip-NeRF360 (single scene)PSNR: 25.2 -> 24.9 dBHurt
Simplify to 0.49M Gaussians with Mini-SplattingMini-Splatting-D · Mip-NeRF360rendering FPS (outdoor): 98 -> 410Improved
Apply depth-based reinitialization on scenes with large skyMini-Splatting · Tanks&TemplesPSNR: 23.66 -> 23.18Hurt
Initialize with dense MVS points instead of sparse SfM (ablation)3DGS · Mip-NeRF360LPIPS: 0.216 -> 0.180Improved
Dense initialization for Mini-Splatting-D (ablation)Mini-Splatting-D · Mip-NeRF360PSNR: 27.54 -> 27.76No clear effect

Later papers that compared with this method

14 later accepted papers report Mini-Splatting as a baseline in their results tables; 13 of them report a better number on most of the results they share with it.

6 more not listed.

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy