MFSR: MeanFlow Distillation for One Step Real-World Image Super Resolution

Ruiqing Wang; Kai Zhang; Yuanzhi Zhu; Hanshu Yan; Shilin Lu; Jian Yang

cs.CVMar 21, 2026

MFSR: MeanFlow Distillation for One Step Real-World Image Super Resolution

Authors: Ruiqing Wang, Kai Zhang, Yuanzhi Zhu, Hanshu Yan, Shilin Lu, Jian Yang

Organizations: Nanjing University · LIX, École Polytechnique, CNRS, IPP · Salesforce AI Research · Nanyang Technological University

Abstract

Diffusion- and flow-based models have advanced Real-world Image Super-Resolution (Real-ISR), but their multi-step sampling makes inference slow and hard to deploy. One-step distillation alleviates the cost, yet often degrades restoration quality and removes the option to refine with more steps. We present Mean Flows for Super-Resolution (MFSR), a new distillation framework that produces photorealistic results in a single step while still allowing an optional few-step path for further improvement. Our approach uses MeanFlow as the learning target, enabling the student to approximate the average velocity between arbitrary states of the Probability Flow ODE (PF-ODE) and effectively capture the teacher's dynamics without explicit rollouts. To better leverage pretrained generative priors, we additionally improve original MeanFlow's Classifier-Free Guidance (CFG) formulation with teacher CFG distillation strategy, which enhances restoration capability and preserves fine details. Experiments on both synthetic and real-world benchmarks demonstrate that MFSR achieves efficient, flexible, and high-quality super-resolution, delivering results on par with or even better than multi-step teachers while requiring much lower computational cost.

Figures & tables

Figure 1: We present MFSR, a high-capacity one- or few-step super-resolution model that delivers photorealistic restoration of real-world low-resolution images. The number of diffusion inference steps is indicated by ‘s’.

Figure 2: Overview of the MeanFlow Distillation (MFD) pipeline. Left: Training pipeline where the student model is initialized from the teacher. Student model needs two time steps as input. Right: Derivation of the MFD loss. The dotted orange line denotes the teacher’s and student’s PF-ODE, the blue line shows the teacher’s predicted instantaneous velocity, the solid orange line shows the student’s predicted average velocity and the purple line indicates the student’s predicted average velocity over a shorter interval. Taking the limit $\Delta t\!\to\!0$ yields the MFD loss $\mathcal{L}_{\text{MFD}}$ .

Datasets	Metric	OSEDiff-1s	AddSR-1s	SinSR-1s	CTMSR-1s	S3Diff-1s	TSDSR-1s	MFSR-1s	MFSR-2s
DRealSR	PSNR $\uparrow$	27.92	27.77	28.27	28.66	27.53	26.19	24.15	24.29
	LPIPS $\downarrow$	0.2966	0.3196	0.3730	0.3232	0.3109	0.3116	0.3660	0.3689
	FID $\downarrow$	135.39	150.18	182.28	162.29	118.49	130.70	143.12	139.56
	NIQE $\downarrow$	6.4381	6.9321	7.0246	6.1426	6.2142	5.7643	6.0241	6.2711
	MUSIQ $\uparrow$	64.67	60.85	55.55	59.84	63.94	66.11	64.47	64.45
	MANIQA $\uparrow$	0.5898	0.5490	0.4907	0.4865	0.6124	0.5820	0.6148	0.6354

Table 1: Quantitative comparison with the state-of-the-art one-step methods across both synthetic and real-world benchmarks. The number of diffusion inference steps is indicated by ‘s’. The best and second best results of each metric are highlighted in red and blue , respectively. S3Diff uses CFG during inference, which actually has 2 number of function evaluations (NFEs).

Figure 3: Qualitative comparison with state-of-the-art methods. All methods perform 1-step inference. Our MFSR is capable of generating vivid details without artifacts or remaining degradations.

Panel of Figure 5 — Figure 5: Qualitative comparison across different sampling steps and our teacher model, DiT4SR. The number of diffusion inference steps is indicated by ’s’.

Panel of Figure 6 — Figure 6: Visual comparison from ablation study of our CFG strategy.

Instantaneous Velocity	$w$	PSNR $\uparrow$	LPIPS $\downarrow$	DISTS $\downarrow$	FID $\downarrow$	NIQE $\downarrow$	MUSIQ $\uparrow$	MANIQA $\uparrow$	CLIPIQA $\uparrow$
$z_{1}-z_{0}$	-	24.21	0.3210	0.2276	121.78	5.5784	65.32	0.6035	0.6474
Original MeanFlow CFG	6	21.20	0.3478	0.2453	120.81	5.8691	67.33	0.6091	0.6951
Ours null	6	25.38	0.2931	0.2151	109.75	5.7146	65.27	0.6184	0.6551
Ours neg	1	24.49	0.2983	0.2237	109.23	5.2234	66.84	0.6273	0.6747
Ours neg	4	22.94	0.3021	0.2255	110.83	5.2317	67.02	0.6317	0.6773
Ours neg	6	21.51	0.3158	0.2295	110.14	5.2421	67.95	0.6389	0.6968

Table 3: Ablation studies for CFG strategy and its scale

w

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

$r$	NIQE $\downarrow$	MUSIQ $\uparrow$	MANIQA $\uparrow$	CLIPIQA $\uparrow$
0	5.2529	67.23	0.6283	0.6819
0.25	5.2311	67.58	0.6301	0.6863
0.5	5.2421	67.95	0.6389	0.6968
0.75	5.1046	67.76	0.6358	0.6895

Appendix

Table 4: Ablation studies for hyperparameter ratio

r

Figure 8: The LR images used in user study.

Panel of Figure 9 — Figure 9: Qualitative comparison across different sampling steps and our teacher model, DiT4SR. The number of diffusion inference steps is indicated by ’s’.

Panel of Figure 10 — Figure 10: Qualitative comparison with state-of-the-art methods. All methods perform 1-step inference. Please zoom in for a better view.

Panel of Figure 11 — Figure 11: Qualitative comparison with state-of-the-art methods. All methods perform 1-step inference. Please zoom in for a better view.

Figure 12: Qualitative comparison with state-of-the-art methods. All methods perform 1-step inference. Please zoom in for a better view.

Panel of Figure 13 — Figure 13: 4 $\times$ SR results on AI-Generated Content using 3-step sampling.

Figure 14: Result of old photo restoration using 3-step sampling.

Image Super-Resolution

Image super-resolution reconstructs high-resolution images from lower-resolution observations, balancing detail, fidelity, and computational cost.

See all 248 papers

RFMSR: Residual Flow Matching for Image Super-Resolution
Jul 14, 2026Shuwei Huang, Tianyao Luo, Jicheng Liu +2Flow Matching Image Super-Resolution
Huazhong University of Science and Technology · Wuhan University
Noise-Started One-Step Real-World Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement
May 10, 2026Wei Zhu, Kai Zhang, Yu Zheng +3Flow Matching Diffusion Model Distillation
Nanjing University of Science and Technology · Nanjing University · Huawei
Fast Image Super-Resolution via Consistency Rectified Flow
May 12, 2026Jiaqi Xu, Wenbo Li, Haoze Sun +8Rectified Flow Consistency Distillation
The Chinese University of Hong Kong · Huawei Noah’s Ark Lab · HKUST (GZ) +1

Jul 14, 2026J/K move · Enter open · S save

RFMSR: Residual Flow Matching for Image Super-Resolution

Shuwei Huang, Tianyao Luo, Jicheng Liu, Daizong Liu+1

Huazhong University of Science and Technology · Wuhan University

Flow Matching Image Super-Resolution

Open paper AskPDF arXiv

Image super-resolution (ISR) has witnessed remarkable progress with diffusion models and flow matching. The dominant text-to-image (T2I) based approaches leverage large-scale foundation models as generative priors, achieving impressive perceptual quality but at the cost of massive model sizes and prohibitive training expenses. Recent flow-matching-based vision-only approaches have made significant strides; however, they adopt standard flow formulations that transport from a pure Gaussian prior to the data distribution, discarding the rich structural information already present in the low-quality (LQ) input. Furthermore, existing single-step acceleration techniques often forfeit the model's multi-step inference capability. In this paper, we propose Residual Flow Matching for Image Super-Resolution (RFMSR), a vision-only framework that centers the source distribution at the LQ latent, reducing transport distance and preserving structural priors throughout the flow trajectory. We further introduce a two-phase training strategy: Phase I pretrains the velocity field via conditional flow matching, while Phase II applies end-to-end supervision to the single-step prediction while retaining the velocity loss across all timesteps, achieving high-quality single-step generation without sacrificing multi-step refinement. Extensive experiments demonstrate that RFMSR achieves comparable or even superior perceptual quality compared to state-of-the-art (SOTA) methods. The source code is available at https://github.com/Faze-Hsw/RFMSR.

Abstract

Image Super-Resolution

Explore similar work

RFMSR: Residual Flow Matching for Image Super-Resolution

Noise-Started One-Step Real-World Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement

Fast Image Super-Resolution via Consistency Rectified Flow