Research map / 3D vision / Depth and stereo
Self-Calibrating Photometric Stereo by Neural Inverse Rendering
Junxuan Li, Hongdong Li
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 progressive specular bases (fixed roughness) (ablation) | Ours · DiLiGenT | normal MAE (deg): 9.39 -> 7.71 | Improved |
| add progressive specular bases (fixed roughness) (ablation) | Ours · DiLiGenT | light direction MAE (deg): 5.30 -> 4.42 | Improved |
| add progressive specular bases (trainable roughness) (ablation) | Ours · DiLiGenT | normal MAE (deg): 8.57 -> 7.05 | Improved |
| make roughness terms trainable (with PSB) (ablation) | Ours · DiLiGenT | normal MAE (deg): 7.71 -> 7.05 | Improved |
| make roughness terms trainable (no PSB) (ablation) | Ours · DiLiGenT | light direction MAE (deg): 5.30 -> 4.75 | Improved |
| add progressive specular bases (trainable roughness) (ablation) | Ours · DiLiGenT | light intensity relative error: 0.0372 -> 0.0365 | Improved |
| perturb initial light directions by up to 70 degrees (ablation) | Ours · DiLiGenT | normal MAE (deg): 7.05 -> 7.44 | No clear effect |
| perturb initial light directions by up to 70 degrees (ablation) | Ours · DiLiGenT | light direction MAE (deg): 4.02 -> 4.40 | No clear effect |
| remove lighting network, use random light initialization (ablation) | Ours · DiLiGenT | light direction MAE (deg): 55.47 degrees | Hurt |
| reduce input to 16 images (ablation) | Ours · DiLiGenT | normal MAE (deg): 7.05 -> 7.77 | Hurt |
| reduce input to 16 images (ablation) | GCNet+PS-FCN · DiLiGenT | normal MAE (deg): 8.70 -> 10.74 | Hurt |
Closest papers by meaning
- Neural Multi-View Self-Calibrated Photometric Stereo without Photometric Stereo Cues (ICCV 2025 · similarity 0.93)
- Uncalibrated Neural Inverse Rendering for Photometric Stereo of General Surfaces (CVPR 2021 · similarity 0.93)
- PS-NeRF: Neural Inverse Rendering for Multi-View Photometric Stereo (ECCV 2022 · similarity 0.92)
- What is Learned in Deep Uncalibrated Photometric Stereo? (ECCV 2020 · similarity 0.91)
- DeepPS2: Revisiting Photometric Stereo Using Two Differently Illuminated Images (ECCV 2022 · similarity 0.91)
- Sparse Views Near Light: A Practical Paradigm for Uncalibrated Point-light Photometric Stereo (CVPR 2024 · similarity 0.91)
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.
