Research map / 3D vision / Gaussian splatting
SmartSplat: Feature-Smart Gaussians for Scalable Compression of Ultra-High-Resolution Images
Linfei Li, Lin Zhang, Zhong Wang, Ying Shen
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 |
|---|---|---|---|
| SmartSplat feature-aware sampling vs 3DGS at same CR | SmartSplat · DIV8K | PSNR: 1.53 dB PSNR gain over 3DGS | Improved |
| SmartSplat vs GI (RS), CR 20 | SmartSplat · DIV8K | PSNR: 30.45 -> 33.26 | Improved |
| SmartSplat vs ImageGS at CR 200 | SmartSplat · DIV8K (CR 200) | PSNR: 26.80 vs 25.75 | Hurt |
| SmartSplat vs ImageGS at CR 20-100 | SmartSplat · DIV8K | PSNR: 32.00 -> 33.26 (CR 20) | Improved |
| SmartSplat vs GI (RS) on 16K | SmartSplat · DIV16K | PSNR: about 5.64 dB gain | Improved |
| SmartSplat still trains at extreme CR where GI and ImageGS fail | SmartSplat · DIV16K (CR 3000) | PSNR: 24.72 (baselines Fail/OOM) | Improved |
| train SmartSplat for only 1K instead of 10K iterations (ablation) | SmartSplat · 10848x16320 image, CR 200 | PSNR: 31.87 -> 30.52 | Mixed |
| SmartSplat 1K iterations vs GI 10K iterations | SmartSplat · 10848x16320 image, CR 200 | PSNR: 19.86 -> 30.52 | Improved |
| SmartSplat vs 3DGS memory and time | SmartSplat · 10848x16320 image, CR 200 | training time (s): 7841.80 -> 2237.52 | Improved |
Closest papers by meaning
- GaussianSR: High Fidelity 2D Gaussian Splatting for Arbitrary-Scale Image Super-Resolution (AAAI 2025 · similarity 0.89)
- Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis (CVPR 2024 · similarity 0.88)
- CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction (NeurIPS 2026 · similarity 0.87)
- CodecSplat: Ultra-Compact Latent Coding for Feed-Forward 3D Gaussian Splatting (NeurIPS 2026 · similarity 0.87)
- S2Gaussian: Sparse-View Super-Resolution 3D Gaussian Splatting (CVPR 2025 · similarity 0.87)
- Large Images Are Gaussians: High-Quality Large Image Representation with Levels of 2D Gaussian Splatting (AAAI 2025 · similarity 0.86)
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.
