Research map / 3D vision / Depth and stereo
Physically-Plausible Illumination Distribution Estimation
Egor Ershov, Vasily Tesalin, Ivan Ermakov, Michael S. Brown
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 |
|---|---|---|---|
| train with fast EMD loss instead of L2 loss on distributions (ablation) | Our histogram-input network · Cube++ illumination distribution test | mean EMD: 2.7 -> 2.0 | Improved |
| use a lightweight histogram-input network instead of modified FC4 for distribution prediction | illumination distribution estimation · Cube++ illumination distribution test | mean EMD: 3.0 -> 2.0 | Improved |
| adapt FC4 to predict histograms vs sliding-window MaxRGB | modified FC4 · Cube++ illumination distribution test | mean EMD: 3.1 -> 3.0 | No clear effect |
| white balance with illuminations sampled inside the GT distribution rather than outside | white balance correction · user preference study (100 images) | mean opinion score: inside >= 0.6; outside spread in [0, 0.6] | Improved |
| keypoint-based sampling instead of sliding-window sampling for spatially-varying baselines | Max RGB · Cube++ illumination distribution test | mean EMD: 3.1 -> 3.5 | Hurt |
Closest papers by meaning
- Attentive Illumination Decomposition Model for Multi-Illuminant White Balancing (CVPR 2024 · similarity 0.88)
- Large Scale Multi-Illuminant (LSMI) Dataset for Developing White Balance Algorithm Under Mixed Illumination (ICCV 2021 · similarity 0.88)
- White-Balance First, Adjust Later: Cross-Camera Color Constancy via Vision-Language Evaluation (CVPR 2026 · similarity 0.84)
- Few-Shot Neural Radiance Fields under Unconstrained Illumination (AAAI 2024 · similarity 0.84)
- Bright-NeRF: Brightening Neural Radiance Field with Color Restoration from Low-Light RAW Images (AAAI 2025 · similarity 0.84)
- MetaGS: A Meta-Learned Gaussian-Phong Model for Out-of-Distribution 3D Scene Relighting (NeurIPS 2025 · similarity 0.84)
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.
