AI Paper Atlas

Research map / 3D vision / Gaussian splatting

Octree-GS: Towards Consistent Real-time Rendering with LOD-Structured 3D Gaussians

Kerui Ren, Lihan Jiang, Tao Lu, Mulin Yu, Linning Xu, Zhangkai Ni, Bo Dai

TPAMI 2025 · 3D vision › Gaussian splatting · approach: Regularization and primitive design for 3DGS · 324 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 next-level anchor growing across LOD levels (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)PSNR: 27.64 -> 28.05Improved
add next-level anchor growing across LOD levels (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)storage memory: 99.7M -> 139.6MHurt
add progressive training (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)PSNR: 27.86 -> 28.05Improved
add LOD bias (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)PSNR: 27.85 -> 28.05Improved
add view frequency pruning (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)PSNR: 27.74 -> 28.05Improved
add view frequency pruning (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)storage memory: 244.4M -> 139.6MImproved
add view frequency pruning (ablation)Octree-GS (Our-Scaffold-GS) · Mip-NeRF360 (9 scenes)LPIPS: 0.211 -> 0.214Hurt
apply octree LOD structure to Scaffold-GSScaffold-GS · Mip-NeRF360 (9 scenes)PSNR: 27.90 -> 28.05Improved
apply octree LOD structure to 3D-GSAnchor-3D-GS · Mip-NeRF360 (9 scenes)PSNR: 27.59 -> 27.65No clear effect
apply octree LOD structure to 2D-GSAnchor-2D-GS · Mip-NeRF360 (9 scenes)PSNR: 26.98 -> 27.02No clear effect
use random point initialization instead of SfM pointsScaffold-GS vs Octree-GS · Mip-NeRF360 (9 scenes, random init)PSNR: 25.93 -> 26.41Improved
apply octree LOD structure to Scaffold-GSScaffold-GS · MatrixCity trajectory T1FPS: 6.69 -> 40.91Improved

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.

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