Research map / 3D vision / Gaussian splatting
Compression of 3D Gaussian Splatting with Optimized Feature Planes and Standard Video Codecs
Soonbin Lee, Fangwen Shu, Yago Sanchez, Thomas Schierl, Cornelius Hellge
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 progressive channel-masking training (ablation) | feature-plane 3DGS compression · Mip-NeRF360 | size (MB): 23.45 -> 22.81 | Improved |
| add progressive training (ablation) | feature-plane 3DGS compression · Mip-NeRF360 | PSNR (dB): 27.27 -> 27.40 | Improved |
| add entropy loss L_ent on top of L1 (ablation) | feature-plane 3DGS compression · Mip-NeRF360 | size (MB): 22.81 -> 10.68 | Improved |
| add entropy loss L_ent on top of L1 (ablation) | feature-plane 3DGS compression · Mip-NeRF360 | PSNR (dB): 27.40 -> 27.31 | No clear effect |
| add channel importance bit allocation weights w_c (ablation) | feature-plane 3DGS compression · Mip-NeRF360 | size (MB): 10.68 -> 9.96 | Improved |
| add piecewise-projective contraction (ablation) | feature-plane 3DGS compression · Mip-NeRF360 | size (MB): 9.96 -> 9.78 | Improved |
| apply the proposed feature-plane method with HM codec (ablation) | 3DGS compression · Tank & Temples | size (MB): 27.9 -> 7.46 | Improved |
| apply the proposed method with libx265 codec (ablation) | 3DGS compression · Tank & Temples | size (MB): 29.2 -> 7.79 | Improved |
| apply the proposed method with HM codec (ablation) | 3DGS compression · Tank & Temples | PSNR (dB): 23.71 -> 23.63 | Hurt |
| apply the proposed method with libx265 codec (ablation) | 3DGS compression · Tank & Temples | Enc/Dec time (s): 27.5/0.7 -> 25.3/0.6 | Improved |
| raise Q_step from 2^0 to 2^8 (ablation) | feature-plane compression · Mip-NeRF360 bonsai | feature plane size (MB): 16.6 -> 8.45 | Improved |
| raise Q_step from 2^8 to 2^12 (ablation) | feature-plane compression · Mip-NeRF360 bonsai | feature plane size (MB): 8.45 -> 17.8 | Hurt |
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.
- Scaffold-GS — Scaffold-GS: Structured 3D Gaussians for View-Adaptive Rendering (CVPR 2024) · better on 3 of 3 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
- CodecSplat: Ultra-Compact Latent Coding for Feed-Forward 3D Gaussian Splatting (NeurIPS 2026 · similarity 0.93)
- Optimized Minimal 3D Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis (CVPR 2024 · similarity 0.91)
- Fast Feedforward 3D Gaussian Splatting Compression (ICLR 2025 · similarity 0.91)
- Gaussian Splatting-based Volumetric Video Compression with Sparse 4D Anchors (NeurIPS 2026 · similarity 0.91)
- Temporal Smoothness-Aware Rate-Distortion Optimized 4D Gaussian Splatting (NeurIPS 2025 · similarity 0.90)
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.
