Research map / 3D vision / Gaussian splatting
A^2TG: Adaptive Anisotropic Textured Gaussians for Efficient 3D Scene Representation
Sheng-Chi Hsu, Ting-Yu Yen, Shih-Hsuan Hung, Hung-Kuo Chu
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 gradient-guided texture upscaling (ablation) | A2TG · average over 3 datasets, 1M Gaussians | PSNR: 27.03 -> 27.37 | Improved |
| add gradient-guided texture upscaling (ablation) | A2TG · average over 3 datasets, 50k Gaussians | PSNR: 24.87 -> 25.17 | Improved |
| use anisotropic (non-square) upscaling instead of square-only (ablation) | A2TG · average over 3 datasets, 1M Gaussians | memory (MB): 298.22 -> 286.54 | Improved |
| use anisotropic (non-square) upscaling instead of square-only (ablation) | A2TG · average over 3 datasets, 1M Gaussians | PSNR: 27.38 -> 27.37 | No clear effect |
| use adaptive textures instead of fixed 4x4 textures on every Gaussian | Textured Gaussians* · Mip-NeRF 360, 1M Gaussians | memory (MB): 488 -> 291 | Improved |
| use adaptive textures instead of fixed 4x4 textures at equal memory budget | Textured Gaussians* · Mip-NeRF 360, fixed memory ~200MB | PSNR: 28.37 -> 28.51 | Improved |
| use adaptive textures instead of fixed textures at 100k Gaussians | Textured Gaussians* · Mip-NeRF 360, 100k Gaussians | PSNR: 27.04 -> 26.62 | Hurt |
| add textures to 2DGS | 2DGS*-MCMC · Mip-NeRF 360, 100k Gaussians | PSNR: 26.40 -> 26.62 | Improved |
| remove SH base color or remove textures (qualitative) (ablation) | A2TG · Mip-NeRF 360 scenes | visual quality | Hurt |
Closest papers by meaning
- Textured Gaussians for Enhanced 3D Scene Appearance Modeling (CVPR 2025 · similarity 0.93)
- IBGS: Image-Based Gaussian Splatting (NeurIPS 2025 · similarity 0.91)
- CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction (NeurIPS 2026 · similarity 0.90)
- Spec-Gaussian: Anisotropic View-Dependent Appearance for 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.90)
- LOD-GS: Achieving Levels of Detail using Scalable Gaussian Soup (CVPR 2025 · similarity 0.90)
- GaussianPro: 3D Gaussian Splatting with Progressive Propagation (ICML 2024 · 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.
