Research map / Image generation / Image restoration
Deep Portrait Delighting
Joshua Weir, Junhong Zhao, Andrew Chalmers, Taehyun Rhee
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 shading-offset loss and decoder (ablation) | L_dlt baseline · Testing dataset | RMSE: 0.056 -> 0.046 | Improved |
| add shading-offset loss and decoder (ablation) | L_dlt baseline · Testing dataset | LPIPS: 0.042 -> 0.035 | Improved |
| add shading-offset loss (ablation) | L_dlt baseline · Alt. Lighting | RMSE: 0.040 -> 0.041 | No clear effect |
| add soft-shadow loss (ablation) | baseline + L_off · Alt. Lighting | RMSE: 0.041 -> 0.026 | Improved |
| add soft-shadow loss (ablation) | baseline + L_off · Alt. Lighting | LPIPS: 0.052 -> 0.033 | Improved |
| add soft-shadow loss (ablation) | baseline + L_off · Testing dataset | RMSE: 0.046 -> 0.046 | No clear effect |
| add masked loss (ablation) | baseline + L_off + L_soft · Testing dataset | RMSE: 0.046 -> 0.044 | Improved |
| add masked loss (ablation) | baseline + L_off + L_soft · Testing dataset | LPIPS: 0.035 -> 0.037 | Hurt |
| add masked loss (ablation) | baseline + L_off + L_soft · Alt. Lighting | LPIPS: 0.033 -> 0.037 | Hurt |
| add masked loss (ablation) | baseline + L_off + L_soft · Testing dataset | SSIM: 0.949 -> 0.946 | Hurt |
Closest papers by meaning
- Deep Single-Image Portrait Relighting (ICCV 2019 · similarity 0.86)
- A Deep Network Solution for Attention and Aesthetics Aware Photo Cropping (TPAMI 2018 · similarity 0.85)
- Composition-Preserving Deep Photo Aesthetics Assessment (CVPR 2016 · similarity 0.85)
- Controllable Light Diffusion for Portraits (CVPR 2023 · similarity 0.85)
- Relightful Harmonization: Lighting-aware Portrait Background Replacement (CVPR 2024 · similarity 0.84)
- SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Faces (CVPR 2025 · similarity 0.83)
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.
