Research map / 3D vision / Gaussian splatting
LiteGS: a high-performance framework to train 3dgs in subminutes via system and algorithm codesign
Kaimin Liao, Hua Wang, Zhi Chen, Luchao Wang, Yaohua Tang
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 Cluster-Cull-Compact (ablation) | LiteGS-quality · Mip-NeRF 360 | Training time (s): 701 -> 515 | Improved |
| add Cluster-Cull-Compact (ablation) | LiteGS-quality · Mip-NeRF 360 | PSNR (dB): 28.28 -> 28.25 | No clear effect |
| add Cluster-Cull-Compact (ablation) | LiteGS-quality · Deep Blending | Training time (s): 533 -> 330 | Improved |
| warp-based rasterizer instead of atomic raster (ablation) | LiteGS · garden scene backward pass | Backward time (ms): 51.84 -> 6.93 | Improved |
| warp-based rasterizer instead of per-splat raster (ablation) | LiteGS · garden scene backward pass | Backward time (ms): 27.31 -> 6.93 | Improved |
| add opacity decay instead of hard reset (ablation) | LiteGS · Mip-NeRF 360 | PSNR (dB): 27.68 -> 28.12 | Improved |
| add opacity-gradient variance densification metric (ablation) | LiteGS · Mip-NeRF 360 | PSNR (dB): 27.40 -> 28.12 | Improved |
| add both opacity decay and variance metric (ablation) | LiteGS · Mip-NeRF 360 | PSNR (dB): 27.13 -> 28.12 | Improved |
| add opacity decay and variance metric (ablation) | LiteGS · Deep Blending | PSNR (dB): 29.76 -> 30.37 | Improved |
| LiteGS-quality vs 3DGS-MCMC | LiteGS · Mip-NeRF 360 | PSNR (dB): 28.08 -> 28.25 | Improved |
| LiteGS-turbo vs Mini-Splatting v2 | LiteGS · Mip-NeRF 360 | Training time (s): 214 -> 145 | Improved |
| LiteGS-turbo vs Mini-Splatting v2 on perceptual quality | LiteGS · Mip-NeRF 360 | LPIPS: 0.215 -> 0.236 | 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.
- Mini-Splatting v2 — Mini-Splatting: Representing Scenes with a Constrained Number of Gaussians (ECCV 2024) · better on 4 of 4 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
- Steepest Descent Density Control for Compact 3D Gaussian Splatting (CVPR 2025 · similarity 0.89)
- Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis (CVPR 2024 · similarity 0.89)
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.89)
- On Scaling Up 3D Gaussian Splatting Training (ICLR 2025 · similarity 0.89)
- Gaussian Herding across Pens: An Optimal Transport Perspective on Global Gaussian Reduction for 3DGS (NeurIPS 2025 · similarity 0.88)
- FlashGS: Efficient 3D Gaussian Splatting for Large-scale and High-resolution Rendering (CVPR 2025 · similarity 0.88)
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.
