Research map / Image generation / Image restoration
Learning Image-adaptive 3D Lookup Tables for High Performance Photo Enhancement in Real-time
Hui Zeng, Jianrui Cai, Lida Li, Zisheng Cao, Lei Zhang
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 CNN weight predictor to a single 3D LUT (ablation) | 3D LUT · FiveK ablation (480p, paired) | PSNR: 20.37 -> 23.15 | Improved |
| increase number of basis LUTs from 1 to 3 (ablation) | 3D LUT + CNN · FiveK ablation (480p, paired) | PSNR: 23.15 -> 25.21 | Improved |
| increase number of basis LUTs from 3 to 5 (ablation) | 3D LUT + CNN · FiveK ablation (480p, paired) | PSNR: 25.21 -> 25.29 | No clear effect |
| increase smoothness weight lambda_s above 0.0001 (ablation) | 3D LUT + CNN · FiveK ablation (480p, paired) | PSNR | Hurt |
| vary monotonicity weight lambda_m (ablation) | 3D LUT + CNN · FiveK ablation (480p, paired) | PSNR | No clear effect |
| add smooth and monotonicity regularization to LUT learning (ablation) | 3D LUT + CNN · FiveK ablation (480p, paired) | banding artifacts (qualitative) | Improved |
| use 3D LUT method instead of HDRNet | Ours vs HDRNet · FiveK photo retouching, paired (480p) | PSNR: 24.32 -> 25.21 | Improved |
| use 3D LUT method instead of HDRNet | Ours vs HDRNet · FiveK photo retouching, paired (original resolution) | PSNR: 24.03 -> 25.10 | Improved |
| use 3D LUT method instead of UIE | Ours vs UIE · FiveK photo retouching, unpaired (480p) | PSNR: 22.11 -> 22.86 | Improved |
| use 3D LUT method instead of HDRNet | Ours vs HDRNet · HDR+ imaging pipeline, paired (480p) | PSNR: 23.04 -> 23.54 | Improved |
| use 3D LUT method instead of UIE | Ours vs UIE · FiveK imaging pipeline, unpaired (480p) | PSNR: 17.83 -> 21.60 | Improved |
| compare to Camera Raw automatic adjustment | Ours vs Camera Raw · FiveK imaging pipeline, unpaired (480p) | PSNR: 21.61 -> 21.60 | No clear effect |
Closest papers by meaning
- SepLUT: Separable Image-Adaptive Lookup Tables for Real-Time Image Enhancement (ECCV 2022 · similarity 0.92)
- NILUT: Conditional Neural Implicit 3D Lookup Tables for Image Enhancement (AAAI 2024 · similarity 0.91)
- Real-Time Exposure Correction via Collaborative Transformations and Adaptive Sampling (CVPR 2024 · similarity 0.91)
- Efficient Neural Network Encoding for 3D Color Lookup Tables (AAAI 2025 · similarity 0.88)
- GLUT: 3D Gaussian Lookup Table for Continuous Color Transformation (NeurIPS 2026 · similarity 0.86)
- Luminance-Aware Statistical Quantization: Unsupervised Hierarchical Learning for Illumination Enhancement (NeurIPS 2025 · similarity 0.85)
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.
