Research map / 3D vision / Gaussian splatting
DiffSoup: Direct Differentiable Rasterization of Triangle Soup for Extreme Radiance Field Simplification
Kenji Tojo, Bernd Bickel, Nobuyuki Umetani
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 learned binary opacity to triangle textures (ablation) | DiffSoup · MipNeRF360 (15K primitives) | PSNR: 24.65 -> 24.76 | Improved |
| add learned binary opacity to triangle textures (ablation) | DiffSoup · MipNeRF360 (15K primitives) | SSIM: 0.744 -> 0.748 | Improved |
| add learned binary opacity to triangle textures (ablation) | DiffSoup · MipNeRF360 (15K primitives) | LPIPS: 0.205 -> 0.204 | No clear effect |
| add multi-resolution feature grid instead of highest-resolution grid only (ablation) | DiffSoup · MipNeRF360 (15K primitives) | PSNR: 23.54 -> 24.48 | Improved |
| add coarse-to-fine optimization on top of multi-resolution grid (ablation) | DiffSoup · MipNeRF360 (15K primitives) | PSNR: 24.48 -> 24.76 | Improved |
| add coarse-to-fine optimization on top of multi-resolution grid (ablation) | DiffSoup · MipNeRF360 (15K primitives) | LPIPS: 0.203 -> 0.204 | No clear effect |
| use DiffSoup instead of 3DGS at 15K primitives | DiffSoup · MipNeRF360 | PSNR: 23.72 -> 24.76 | Improved |
| use DiffSoup instead of TexGS at 15K primitives | DiffSoup · MipNeRF360 | LPIPS: 0.270 -> 0.204 | Improved |
| use DiffSoup instead of TexGS at 15K primitives | DiffSoup · MipNeRF360 | PSNR: 24.80 -> 24.76 | No clear effect |
| use DiffSoup CUDA rendering instead of 3DGS | DiffSoup · MipNeRF360 full resolution | FPS: 115 -> 1.96K | Improved |
| initialize from QEM-decimated MobileNeRF instead of random initialization | DiffSoup · NeRF-Synthetic ship | PSNR: 25.29 -> 26.68 | Improved |
| initialize from QEM-decimated MobileNeRF instead of random initialization | DiffSoup · Shelly khady | PSNR: 26.67 -> 26.55 | No clear effect |
Closest papers by meaning
- Compact 3D Gaussian Representation for Radiance Field (CVPR 2024 · similarity 0.90)
- 3D Convex Splatting: Radiance Field Rendering with 3D Smooth Convexes (CVPR 2025 · similarity 0.89)
- HyRF: Hybrid Radiance Fields for Memory-efficient and High-quality Novel View Synthesis (NeurIPS 2025 · similarity 0.89)
- pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction (CVPR 2024 · similarity 0.89)
- DNGaussian: Optimizing Sparse-View 3D Gaussian Radiance Fields with Global-Local Depth Normalization (CVPR 2024 · similarity 0.88)
- Sparse Voxels Rasterization: Real-time High-fidelity Radiance Field 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.
