Research map / Image generation / Super-resolution
FiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution
Aro Kim, Myeongjin Jang, Chaewon Moon, Youngjin Shin, Jinwoo Jeong, Sang-hyo Park
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 LRRB and DAW to the one-step baseline (ablation) | FiDeSR · DIV2K | MUSIQ: 67.6043 -> 68.2869 | Improved |
| add LRRB (latent residual refinement) (ablation) | FiDeSR without DAW · DIV2K | MANIQA: 0.6278 -> 0.6285 | No clear effect |
| add DAW (ablation) | FiDeSR without LRRB · DIV2K | MANIQA: 0.6236 -> 0.6278 | No clear effect |
| add LRRB to baseline (ablation) | baseline · HF noise prediction MSE (avg) | MSE: 0.1048 -> 0.1032 | Improved |
| inject high-frequency components (strength 0.1 to 0.5) (ablation) | FiDeSR · RealSR | MANIQA / PSNR: MANIQA 0.6580 -> 0.6803; PSNR 26.2542 -> 25.5778 | Mixed |
| inject low-frequency components (strength 0.1 to 0.5) (ablation) | FiDeSR · RealSR | PSNR / MANIQA: PSNR 26.2542 -> 26.3448; MANIQA 0.6580 -> 0.6522 | Mixed |
| use FiDeSR rather than PiSA-SR | PiSA-SR · RealSR | FID: 124.18 -> 109.68 | Improved |
| use FiDeSR rather than PiSA-SR on DIV2K | PiSA-SR · DIV2K | MANIQA: 0.6375 -> 0.6384 | No clear effect |
| use FiDeSR rather than PiSA-SR on DRealSR | PiSA-SR · DRealSR | PSNR: 28.32 -> 28.90 | Improved |
Closest papers by meaning
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.92)
- Arbitrary-steps Image Super-resolution via Diffusion Inversion (CVPR 2025 · similarity 0.91)
- RAP-SR: RestorAtion Prior Enhancement in Diffusion Models for Realistic Image Super-Resolution (AAAI 2025 · similarity 0.91)
- Taming Diffusion Prior for Image Super-Resolution with Domain Shift SDEs (NeurIPS 2024 · similarity 0.90)
- TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution (CVPR 2025 · similarity 0.90)
- Noise-Started One-Step Image Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement (NeurIPS 2026 · 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.
