Research map / 3D vision / Face and shape
Learning Image Harmonization in the Linear Color Space
Ke Xu, Gerhard Petrus Hancke, Rynson W.H. Lau
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 image unprocessing to convert to linear color space (ablation) | color harmonization network · iHarmony4 average (256x256) | PSNR: 37.79 -> 39.36 | Improved |
| use diffusion process in image unprocessing instead of standard convolution (ablation) | image unprocessing · iHarmony4 average (256x256) | PSNR: 38.28 -> 39.36 | Improved |
| add feature harmonization (bilateral propagation) (ablation) | color harmonization · iHarmony4 average (256x256) | PSNR: 38.10 -> 39.36 | Improved |
| add background preserving guided rendering (ablation) | color rendering · iHarmony4 average (256x256) | PSNR: 39.17 -> 39.36 | Improved |
| use two encoders instead of a single encoder (ablation) | color harmonization network · iHarmony4 average (256x256) | PSNR: 39.02 -> 39.36 | Improved |
| add SE block in feature harmonization (ablation) | feature harmonization · iHarmony4 average (256x256) | PSNR: 38.47 -> 39.36 | Improved |
| add spatial attention in feature harmonization (ablation) | feature harmonization · iHarmony4 average (256x256) | PSNR: 38.19 -> 39.36 | Improved |
| fine-tune pre-trained DoveNet in linear space | DoveNet · iHarmony4 | PSNR: 34.75 -> 33.18 (-1.57) | Hurt |
| train DoveNet from scratch in linear space | DoveNet · iHarmony4 | PSNR: +0.19 (34.94) | No clear effect |
| use ControlNet diffusion model for harmonization | ControlNet · iHarmony4 | PSNR: PSNR/MAE 14.36/109.40 | Hurt |
| learned image unprocessing module vs prior sRGB-to-linear methods | image unprocessing · Adobe5K CIE XYZ | PSNR: 29.66 -> 30.34 | Improved |
Closest papers by meaning
- Progressive Painterly Image Harmonization from Low-Level Styles to High-Level Styles (AAAI 2024 · similarity 0.87)
- High-Resolution Image Harmonization with Adaptive-Interval Color Transformation (NeurIPS 2024 · similarity 0.87)
- Relightful Harmonization: Lighting-aware Portrait Background Replacement (CVPR 2024 · similarity 0.87)
- Painterly Image Harmonization by Learning from Painterly Objects (AAAI 2024 · similarity 0.86)
- IReNe: Instant Recoloring of Neural Radiance Fields (CVPR 2024 · similarity 0.86)
- Training-and-Prompt-Free General Painterly Harmonization via Zero-Shot Disentenglement on Style and Content References (AAAI 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.
