Research map / Image generation / Text-to-image diffusion
HDR Image Generation via Gain Map Decomposed Diffusion
Yuanshen Guan, Ruikang Xu, Yinuo Liao, Mingde Yao, Lizhi Wang, Zhiwei Xiong
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 brightness-aware compression instead of hard clipping (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | FHLP: 0.0036 -> 0.1009 | Improved |
| add brightness-aware compression instead of hard clipping (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | NIQE: 8.1674 -> 6.4392 | Improved |
| add brightness-aware compression instead of fixed mu-law (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | EHL: 0.5897 -> 0.7051 | Improved |
| add brightness-aware compression instead of fixed mu-law (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | FHLP: 0.0855 -> 0.1009 | Improved |
| add gamut-constrained reduction (gamut loss) (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | FWGP: 0.0010 -> 0.0058 | Improved |
| add gamut-constrained reduction (gamut loss) (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | NIQE: 7.2908 -> 6.4392 | Improved |
| add gamut-constrained reduction (gamut loss) (ablation) | triplet construction · MSCOCO subset (5,000 samples), degradation ablation | BRISQUE: 30.6521 -> 27.2434 | Improved |
| use gain map diffusion instead of SD+KUNet pipeline | HDR image generation · HDR generation (4,000 landscape samples) | EHL: 0.0575 -> 0.0900 | Improved |
| use gain map diffusion instead of SD+KUNet pipeline | HDR image generation · HDR generation (4,000 landscape samples) | FHLP: 0.5764 -> 0.6318 | Improved |
| use gain map diffusion instead of KUNet | SDR-to-HDRTV up-conversion · HDRTV1K SDR-to-HDRTV up-conversion | EWG: 0.0006 -> 0.0295 | Improved |
| use gain map diffusion instead of HDRTVNet | SDR-to-HDRTV up-conversion · HDRTV1K SDR-to-HDRTV up-conversion | HDRVDP3: 8.4642 -> 7.4771 | Hurt |
| use gain map diffusion instead of best baseline | SDR-to-HDRTV up-conversion · HDRTV1K SDR-to-HDRTV up-conversion | luminance range extension and color gamut coverage: 3.77× and 6.02× | Improved |
Closest papers by meaning
- Single Shot HDR Recovery via a Video Diffusion Prior (NeurIPS 2026 · similarity 0.90)
- GaussHDR: High Dynamic Range Gaussian Splatting via Learning Unified 3D and 2D Local Tone Mapping (CVPR 2025 · similarity 0.88)
- Generating Content for HDR Deghosting from Frequency View (CVPR 2024 · similarity 0.88)
- Zero-Shot Structure-Preserving Diffusion Model for High Dynamic Range Tone Mapping (CVPR 2024 · similarity 0.88)
- HDR-GS: Efficient High Dynamic Range Novel View Synthesis at 1000x Speed via Gaussian Splatting (NeurIPS 2024 · similarity 0.88)
- LEDiff: Latent Exposure Diffusion for HDR Generation (CVPR 2025 · similarity 0.87)
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.
