Research map / 3D vision / Depth and stereo
EMLight: Lighting Estimation via Spherical Distribution Approximation
Fangneng Zhan, Changgong Zhang, Yingchen Yu, Yuan Chang, Shijian Lu, Feiying Ma, Xuansong Xie
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 |
|---|---|---|---|
| replace spherical Gaussian representation with spherical distribution representation (ablation) | EMLight regression network · Laval Indoor HDR test, diffuse | RMSE: 0.204 -> 0.133 | Improved |
| replace L2 loss with spherical mover's loss (SML) (ablation) | EMLight spherical-distribution regression · Laval Indoor HDR test, diffuse | RMSE: 0.133 -> 0.080 | Improved |
| add the neural projector (ablation) | EMLight with spherical distribution and SML · Laval Indoor HDR test, diffuse | RMSE: 0.080 -> 0.062 | Improved |
| replace SML with cross-entropy loss (ablation) | EMLight* · Laval Indoor HDR test | RMSE: 0.074 -> 0.102 | Hurt |
| replace spherical convolution with normal convolution (ablation) | EMLight* · Laval Indoor HDR test | RMSE: 0.074 -> 0.086 | Hurt |
| reduce anchor points from 128 to 64 (ablation) | EMLight* · Laval Indoor HDR test | RMSE: 0.074 -> 0.091 | Hurt |
| increase anchor points from 128 to 196 (ablation) | EMLight* · Laval Indoor HDR test | si-RMSE: 0.058 -> 0.055 | Improved |
| increase anchor points from 128 to 196 (ablation) | EMLight* · Laval Indoor HDR test | RMSE: 0.074 -> 0.076 | No clear effect |
| use EMLight instead of the strongest reported baseline | lighting estimation · Laval Indoor HDR test, diffuse | RMSE: 0.084 -> 0.062 | Improved |
| use EMLight instead of the strongest reported baseline | lighting estimation · Laval Indoor HDR test, diffuse | AMT: 33.5\% -> 40.0\% | Improved |
Later papers that compared with this method
2 later accepted papers report EMlight as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- StyleLight: HDR Panorama Generation for Lighting Estimation and Editing (ECCV 2022 · better on 10 of 10 shared results)
- Editable Indoor Lighting Estimation (ECCV 2022 · better on 4 of 5 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ROGR: Relightable 3D Objects using Generative Relighting (NeurIPS 2025 · similarity 0.87)
- ReCap: Better Gaussian Relighting with Cross-Environment Captures (CVPR 2025 · similarity 0.87)
- DiffusionLight: Light Probes for Free by Painting a Chrome Ball (CVPR 2024 · similarity 0.87)
- IllumiNeRF: 3D Relighting Without Inverse Rendering (NeurIPS 2024 · similarity 0.87)
- Generative Multiview Relighting for 3D Reconstruction under Extreme Illumination Variation (CVPR 2025 · similarity 0.87)
- Neural LightRig: Unlocking Accurate Object Normal and Material Estimation with Multi-Light Diffusion (CVPR 2025 · similarity 0.86)
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.
