Research map / Image generation / Image restoration
PI-Light: Physics-Inspired Diffusion for Full-Image Relighting
Zhexin Liang, Zhaoxi Chen, Yongwei Chen, Tianyi Wei, Tengfei Wang, Xingang Pan
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 decomposition into intrinsics before relighting (ablation) | PI-Light forward rendering · Forward rendering ablation | PSNR: 10.88 -> 11.95 | Improved |
| add L_DS shading loss on top of decomposition (ablation) | PI-Light forward rendering · Forward rendering ablation | PSNR: 11.95 -> 11.97 | No clear effect |
| add L_PS shading loss on top of decomposition (ablation) | PI-Light forward rendering · Forward rendering ablation | PSNR: 11.95 -> 12.18 | Improved |
| add both L_DS and L_PS (ablation) | PI-Light forward rendering · Forward rendering ablation | PSNR: 11.95 -> 12.68 | Improved |
| increase CFG scale from 1.0 to 2.5 (ablation) | PI-Light inverse rendering · Object50 | average PSNR: 20.53 -> 18.73 | Hurt |
| increase CFG scale from 1.0 to 2.5 (ablation) | PI-Light inverse rendering · Object50 | average LPIPS: 0.0761 -> 0.0852 | Hurt |
| compare PI-Light with RGB<->X retrained on same data | PI-Light · Object50 | average PSNR: 19.38 -> 20.53 | Improved |
| compare PI-Light with Neural Gaffer on shared subset | PI-Light · Shared training subset | PSNR: 10.31 -> 11.74 | Improved |
| compare PI-Light with RGB<->X on Scene200 retrained | PI-Light · Scene200 | average PSNR: 11.70 -> 12.95 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Neural Gaffer — Neural Gaffer: Relighting Any Object via Diffusion (NeurIPS 2024) · better on 2 of 2 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
- Neural Gaffer: Relighting Any Object via Diffusion (NeurIPS 2024 · similarity 0.90)
- UniLumos: Fast and Unified Image and Video Relighting with Physics-Plausible Feedback (NeurIPS 2025 · similarity 0.90)
- SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Faces (CVPR 2025 · similarity 0.90)
- UniRelight: Learning Joint Decomposition and Synthesis for Video Relighting (NeurIPS 2025 · similarity 0.90)
- DiffusionLight: Light Probes for Free by Painting a Chrome Ball (CVPR 2024 · similarity 0.89)
- Diffusion Renderer: Neural Inverse and Forward Rendering with Video Diffusion Models (CVPR 2025 · similarity 0.89)
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.
