Research map / 3D vision / Depth and stereo
Learning High Dynamic Range From Outdoor Panoramas
Jinsong Zhang, Jean-Francois Lalonde
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 sun elevation loss and render loss to HDR loss (ablation) | HDR regression CNN · Synthetic test set | all error metrics | Improved |
| add random color shifts to training data (ablation) | Baseline · Real Theta S panoramas | E_HDR: 5.34 -> 5.32 | No clear effect |
| add camera response functions to training data (ablation) | Color model · Real Theta S panoramas | E_HDR: 5.32 -> 3.59 | Improved |
| add camera response functions to training data (ablation) | Color model · Real Theta S panoramas | E_theta: 0.10 -> 0.08 | Improved |
| add unsupervised domain adaptation (ablation) | Tone model · Real Theta S panoramas | E_HDR: 3.59 -> 7.53 | Hurt |
| fine-tune on real HDR ground truth (ablation) | Tone model · Real Theta S panoramas | E_HDR: 3.59 -> 2.55 | Improved |
| fine-tune on real HDR ground truth (ablation) | Tone model · Real Theta S panoramas | E_render: 1.06 -> 0.64 | Improved |
| linearize input with RF+WB camera model (ablation) | Baseline · Real Theta S panoramas | E_HDR: 5.34 -> 2.99 | Improved |
| train on Tone augmented data when input is linearized with RF+WB (ablation) | Baseline · Real Theta S panoramas | E_HDR: 2.99 -> 6.55 | Hurt |
| apply gamma 2.2 to input JPGs (ablation) | Baseline · Real Theta S panoramas | E_HDR: 5.34 -> 6.63 | Hurt |
| replace inverse tone mapping methods with learned network | iTMO · Real test set | E_HDR: 29.1 -> 2.55 | Improved |
| replace physically based sky model fit with learned network | Hold-Geoffroy sky model · Real test set | E_sun: 1.70 -> 0.22 | Improved |
Closest papers by meaning
- Deep Outdoor Illumination Estimation (CVPR 2017 · similarity 0.90)
- NeRF for Outdoor Scene Relighting (ECCV 2022 · similarity 0.87)
- All-Weather Deep Outdoor Lighting Estimation (CVPR 2019 · similarity 0.87)
- LuxDiT: Lighting Estimation with Video Diffusion Transformer (NeurIPS 2025 · similarity 0.86)
- Pano-NeRF: Synthesizing High Dynamic Range Novel Views with Geometry from Sparse Low Dynamic Range Panoramic Images (AAAI 2024 · similarity 0.85)
- Real-Time Scene-Adaptive Tone Mapping for High-Dynamic Range Object Detection (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.
