AI Paper Atlas

Research map / Benchmarks / DRealSR

DRealSR leaderboard

57 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on DRealSR. Best PSNR: 29.678 (StableSR (Ours), AAAI 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

DRealSR · PSNR

32 papers, each with its best reported number for its own method; higher is better.

#PSNR ↑MethodPaperTopic
129.678StableSR (Ours)
StableSR
Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution
AAAI 2025 · Table 1
Super-resolution
229.5SRSR-SeeSR-s50
SeeSR-s50
SRSR: Enhancing Semantic Accuracy in Real-World Image Super-Resolution with Spatially Re-Focused Text-Conditioning
NeurIPS 2025 · Table 1
Super-resolution
329.33BiMaCoSR
SinSR
BiMaCoSR: Binary One-Step Diffusion Model Leveraging Flexible Matrix Compression for Real Super-Resolution
ICML 2025 · Table 1
Quantization
429.0948PASD
Stable Diffusion
Pixel-Aware Stable Diffusion for Realistic Image Super-Resolution and Personalized Stylization
ECCV 2024 · Table 1
Super-resolution
529.08FreqOrtho-SR
SD 2.1-base
FreqOrtho-SR: Frequency-Guided Orthogonal Expert Learning for Real-World Image Super-Resolution
ECCV 2026 · Table 1
Super-resolution
628.9RCODO-Fid.
OSEDiff
Realism Control One-step Diffusion for Real-world Image Super Resolution
AAAI 2026 · Table 2
Super-resolution
728.9FiDeSR-1sFiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution
CVPR 2026 · Table 1
Super-resolution
828.77ASDSR (Ours)Adversarial Score Distillation for Stable One-Step Diffusion in Real-World Image Super-Resolution
ECCV 2026 · Table 2
Super-resolution
928.721SwinIR (RealDGen)
SwinIR
Towards Realistic Data Generation for Real-World Super-Resolution
ICLR 2025 · Table: quantitative comparison
Super-resolution
1028.387TADSR (ts=500)
SD 2.1-base
Time-Aware One Step Diffusion Network for Real-World Image Super-Resolution
CVPR 2026 · Table 1
Super-resolution
22 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

DRealSR · MUSIQ

25 papers, each with its best reported number for its own method; higher is better.

15 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

DRealSR · LPIPS

23 papers, each with its best reported number for its own method; lower is better.

#LPIPS ↓MethodPaperTopic
10.2567StableSR (Ours)
StableSR
Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution
AAAI 2025 · Table 1
Super-resolution
20.2592ODTSR (f=1.0, w/o prompt)
Qwen-Image
One-Step Diffusion Transformer for Controllable Real-World Image Super-Resolution
CVPR 2026 · Table 2
Super-resolution
30.2627AdaDiffSR
Stable Diffusion 2.1-base
AdaDiffSR: Adaptive Region-aware Dynamic acceleration Diffusion Model for Real-World Image Super-Resolution
ECCV 2024 · Table 1
Super-resolution
40.2702CoSeR
Stable Diffusion 2.1-base
CoSeR: Bridging Image and Language for Cognitive Super-Resolution
CVPR 2024 · Table 1
Super-resolution
50.2799FreqOrtho-SR
SD 2.1-base
FreqOrtho-SR: Frequency-Guided Orthogonal Expert Learning for Real-World Image Super-Resolution
ECCV 2026 · Table 1
Super-resolution
60.2836FiDeSR-1sFiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution
CVPR 2026 · Table 1
Super-resolution
70.284SFD
RRDB
Exploring Semantic Feature Discrimination for Perceptual Image Super-Resolution and Opinion-Unaware No-Reference Image Quality Assessment
CVPR 2025 · Table 2
Super-resolution
80.2866SRSR-SeeSR-s50
SeeSR-s50
SRSR: Enhancing Semantic Accuracy in Real-World Image Super-Resolution with Spatially Re-Focused Text-Conditioning
NeurIPS 2025 · Table 1
Super-resolution
90.2874ASASR
FLUX.1-dev
Coloring the Noise: Adversarial Sobolev Alignment for Faithful Image Super Resolution
ICML 2026 · Table 1
Super-resolution
100.2896ASDSR (Ours)Adversarial Score Distillation for Stable One-Step Diffusion in Real-World Image Super-Resolution
ECCV 2026 · Table 2
Super-resolution
13 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

DRealSR · CLIPIQA

22 papers, each with its best reported number for its own method; higher is better.

12 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

DRealSR · MANIQA

17 papers, each with its best reported number for its own method; higher is better.

#MANIQA ↑MethodPaperTopic
10.6492ResFlow-Tuner
FLUX.1-dev
Tuning Real-World Image Restoration at Inference: A Test-Time Scaling Paradigm for Flow Matching Models
ECCV 2026 · Table 2
Fast diffusion sampling
20.6394SCR-1Restoration Adaptation for Semantic Segmentation on Low Quality Images
IJCV 2026 · Table 2
Open-set and few-shot detection
30.6317RCODS-Real.
S3Diff
Realism Control One-step Diffusion for Real-world Image Super Resolution
AAAI 2026 · Table 2
Super-resolution
40.627DiT4SR
SD3
DiT4SR: Taming Diffusion Transformer for Real-World Image Super-Resolution
ICCV 2025 · Table 1
Super-resolution
50.6264AlloSR2
FLUX.1-dev
Allo{SR}2: Rectifying One-Step Super-Resolution to Stay Real via Allomorphic Generative Flows
ECCV 2026 · Table 1
Super-resolution
60.6239FiDeSR-1sFiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution
CVPR 2026 · Table 1
Super-resolution
70.622SMFSR
SD3.5-Medium
Noise-Started One-Step Image Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement
NeurIPS 2026 · Table 1
Super-resolution
80.6161SeeSR + RAP-SR, default
SeeSR
RAP-SR: RestorAtion Prior Enhancement in Diffusion Models for Realistic Image Super-Resolution
AAAI 2025 · Ablation study table
Super-resolution
90.6024FluxIR
Flux
Acquire and then Adapt: Squeezing out Text-to-Image Model for Image Restoration
CVPR 2025 · Table 1
Text-to-image diffusion
100.5874Ours-1s
SD3
TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution
CVPR 2025 · Table 1
Super-resolution
7 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

DRealSR · SSIM

15 papers, each with its best reported number for its own method; higher is better.

#SSIM ↑MethodPaperTopic
10.8415AdaDiffSR
Stable Diffusion 2.1-base
AdaDiffSR: Adaptive Region-aware Dynamic acceleration Diffusion Model for Real-World Image Super-Resolution
ECCV 2024 · Table 1
Super-resolution
20.8267StableSR (Ours)
StableSR
Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution
AAAI 2025 · Table 1
Super-resolution
30.815SFD
RRDB
Exploring Semantic Feature Discrimination for Perceptual Image Super-Resolution and Opinion-Unaware No-Reference Image Quality Assessment
CVPR 2025 · Table 2
Super-resolution
40.8128SRSR-SeeSR-s50
SeeSR-s50
SRSR: Enhancing Semantic Accuracy in Real-World Image Super-Resolution with Spatially Re-Focused Text-Conditioning
NeurIPS 2025 · Table 1
Super-resolution
50.7937StableSR + StructSR
StableSR
StructSR: Refuse Spurious Details in Real-World Image Super-Resolution
AAAI 2025 · Table 1
Super-resolution
60.776iRAG (Ours)Reference-based Super-Resolution via Image-based Retrieval-Augmented Generation Diffusion
ICCV 2025 · Table 1
Super-resolution
70.7726AdcSR
OSEDiff
Adversarial Diffusion Compression for Real-World Image Super-Resolution
CVPR 2025 · Table 1
Super-resolution
80.7675PocketSR
SD-Turbo
PocketSR: The Super-Resolution Expert in Your Pocket Mobiles
NeurIPS 2025 · Table 1
Super-resolution
90.7652VARSR with SA-RoPEVisual Autoregressive Modeling for Image Super-Resolution
ICML 2025 · Scale-aligned RoPE ablation
Super-resolution
100.758F2N-ESRGAN (Ours)
ESRGAN
From Face to Natural Image: Learning Real Degradation for Blind Image Super-Resolution
ECCV 2022 · Table 1
Super-resolution
5 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.