Research map / 3D vision / 3D generation
Stanford-ORB: A Real-World 3D Object Inverse Rendering Benchmark
Zhengfei Kuang, Yunzhi Zhang, Hong-Xing Yu, Samir Agarwala, Shangzhe Wu, Jiajun Wu
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 |
|---|---|---|---|
| use IDR (SDF surface) instead of NeRF | object inverse rendering · Stanford-ORB novel view synthesis | PSNR-H: 26.31 -> 30.11 | Improved |
| use IDR instead of NeRF | object inverse rendering · Stanford-ORB geometry | Chamfer distance x1e-3: 62.05 -> 0.30 | Improved |
| extend IDR/NeRF with material and lighting decomposition | IDR/NeRF · Stanford-ORB novel view synthesis | PSNR-H: 30.11 -> 25.91 (InvRender) | Hurt |
| add Monte Carlo differentiable renderer (NVDiffRecMC vs NVDiffRec) | NVDiffRec · Stanford-ORB novel view synthesis | PSNR-H: 21.94 -> 28.03 | Improved |
| add Monte Carlo differentiable renderer (NVDiffRecMC vs NVDiffRec) | NVDiffRec · Stanford-ORB novel scene relighting | PSNR-H: 22.91 -> 24.43 | Improved |
| train with ground-truth 3D scans and pseudo materials | NVDiffRecMC · Stanford-ORB novel scene relighting | PSNR-H: 24.43 -> 25.08 | Improved |
| add Monte Carlo renderer when ground-truth scans are used | NVDiffRec with GT scans · Stanford-ORB novel scene relighting | PSNR-H: 24.93 -> 25.08 | No clear effect |
| change object appearance (dark Salt can vs Baking can of similar shape) | NVDiffRecMC · Stanford-ORB novel scene relighting | PSNR-H: 24.4 -> 7.5 | Hurt |
Closest papers by meaning
- OpenIllumination: A Multi-Illumination Dataset for Inverse Rendering Evaluation on Real Objects (NeurIPS 2023 · similarity 0.88)
- 3DReflecNet: A Large-Scale Dataset for 3D Reconstruction of Reflective, Transparent, and Low-Texture Objects (CVPR 2026 · similarity 0.87)
- ComGS: Efficient 3D Object-Scene Composition via Surface Octahedral Probes (ICLR 2026 · similarity 0.87)
- Inverse Rendering using Multi-Bounce Path Tracing and Reservoir Sampling (ICLR 2025 · similarity 0.86)
- Eclipse: Disambiguating Illumination and Materials using Unintended Shadows (CVPR 2024 · similarity 0.86)
- SVG-IR: Spatially-Varying Gaussian Splatting for Inverse Rendering (CVPR 2025 · similarity 0.85)
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.
