Research map / Image generation / Super-resolution
Exploiting Diffusion Prior for Real-World Image Super-Resolution
Jianyi Wang, Zongsheng Yue, Shangchen Zhou, Kelvin C. K. Chan, Chen Change Loy
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 time-aware guidance (ablation) | StableSR · RealSR | LPIPS: 0.3157 -> 0.3002 | Improved |
| add time-aware guidance (ablation) | StableSR · DRealSR | SSIM: 0.7280 -> 0.7536 | Improved |
| add pixel color correction (ablation) | StableSR · RealSR | PSNR: 22.24 -> 24.65 | Improved |
| add pixel color correction (ablation) | StableSR · DRealSR | LPIPS: 0.3544 -> 0.3284 | Improved |
| use pixel instead of wavelet color correction (ablation) | StableSR · RealSR | PSNR: 23.38 -> 24.65 | Improved |
| raise CFW coefficient w from 0.0 to 0.75 (ablation) | StableSR · DIV2K Valid | PSNR: 22.68 -> 24.17 | Improved |
| raise CFW coefficient w from 0.0 to 0.75 (ablation) | StableSR · RealSR | CLIP-IQA: 0.6127 -> 0.5341 | Hurt |
| raise CFW coefficient w from 0.5 to 1.0 (ablation) | StableSR · DIV2K Valid | FID: 24.44 -> 26.14 | Hurt |
| increase negative-prompt guidance scale s from 0.0 to 7.5 (prompt a) (ablation) | StableSR · DRealSR | MUSIQ: 57.31 -> 62.41 | Improved |
| increase negative-prompt guidance scale s from 0.0 to 7.5 (prompt a) (ablation) | StableSR · RealSR | PSNR: 24.80 -> 23.53 | Hurt |
| add positive prompt (1) without guidance (ablation) | StableSR · RealSR | LPIPS: 0.3002 -> 0.3151 | Hurt |
| use SD-Turbo with 4 steps instead of 200 DDPM steps (ablation) | StableSR · runtime | seconds: 15.16s -> 0.83s | Improved |
Closest papers by meaning
- Taming Diffusion Prior for Image Super-Resolution with Domain Shift SDEs (NeurIPS 2024 · similarity 0.93)
- Unleashing the Power of One-Step Diffusion based Image Super-Resolution via a Large-Scale Diffusion Discriminator (NeurIPS 2025 · similarity 0.92)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.91)
- DREAM: Diffusion Rectification and Estimation-Adaptive Models (CVPR 2024 · similarity 0.91)
- RAP-SR: RestorAtion Prior Enhancement in Diffusion Models for Realistic Image Super-Resolution (AAAI 2025 · similarity 0.91)
- Solving Diffusion ODEs with Optimal Boundary Conditions for Better Image Super-Resolution (ICLR 2024 · similarity 0.90)
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.
