Latest JPEG restoration systems achieve strong quality with large models, yet often remain too slow and expensive for efficient on-device deployment. We present a 65M-parameter generative restorer that attains the lowest LPIPS at QF 10 and 20 on LIVE-1, Urban100, and DIV2K-val while sustaining 8.05 images/s at 1024×1024 on a single RTX 3090, roughly 4.9× the reported throughput of one-step SODiff at one-twentieth of its parameters. Trained from scratch, the model replaces the learned VAE encoder-decoder with an exactly invertible two-level Haar transform, predicts a clean wavelet-domain residual through a rank-enhanced linear-attention DiT that estimates compression severity internally, and is optimized with an improved MeanFlow objective that enables inference in one or two network evaluations without distillation. Large pretrained priors remain stronger under severe compression (QF 5), whereas our model prioritizes throughput for deployment-constrained restoration.
Figures & tables
Figure 1: The quality-throughput frontier on LIVE-1 at QF 10 for 1024×1024 inputs. Two-update MeanFlow approaches SODiff’s LPIPS while moving to a substantially higher-throughput regime; one-update MeanFlow exposes a faster operating point. Throughput uses a logarithmic axis.
Method
Parameters (M)
NFE
Latency (s)
Throughput (img/s)
FBCNN
70.10
1
0.275
3.63
JDEC
38.90
1
0.860
1.16
PromptCIR
34.75
64
27.97
0.0358
SUPIR
4490
50
48.66
0.0206
SODiff
1288
1
0.610
1.64
Ours
65.32
2
0.1242
8.05
Table 3: System-level efficiency for batch-one 1024×1024 restoration.
Configuration
Batch latency (s)
Throughput (img/s)
Peak memory (GiB)
LPIPS
RF BF16, Euler-20
10.6677
1.4999
2.0378
0.177425
MeanFlow BF16, uniform-2
1.2689
12.6096
1.9794
0.160386
+ torch.compile
0.5961
26.8415
0.5107
0.160386
+ INT8 weights
0.6093
26.2585
0.4907
0.160416
Table 5: Algorithmic and implementation changes on LIVE-1 at QF 10 with a batch of 16. Batch latency measures the complete 16-image batch.
Diffusion-based video restoration recovers realistic details, but its practical deployment is limited by two efficiency bottlenecks: costly VAE encoding and decoding, and the quadratic cost of full self-attention in diffusion transformers (DiTs). This paper presents FastVR, a streaming video restoration framework built on a one-step diffusion model, which delivers strong restoration quality and temporal consistency while processing 1080p video at 11 FPS on a single H20 GPU. To improve inference efficiency, FastVR combines a lightweight VAE with chunk-wise causal attention, which substantially reduces the computational cost. During training, it further adopts velocity consistency regularization and continuous trajectory learning, which improve restoration quality. Extensive experiments show that FastVR is more efficient than the evaluated diffusion baselines while achieving state-of-the-art performance on synthetic and real-world benchmarks. We hope that this work supports further progress in the community.
In this paper, we introduce SP3, a novel Plug-and-Play algorithm that accelerates maximum a posteriori image restoration by replacing denoisers with Spherical Encoders (SE) as generative priors. SP3 approximates the intractable proximal prior step by utilizing the SE tightly structured latent space as a robust projection onto the natural image manifold. Alternating this projection with a closed-form data-consistency step, via Half-Quadratic Splitting, achieves stable convergence without requiring gradient computation during inference. This unique formulation unlocks "anytime" restoration capabilities, producing sharp, plausible images from the first iteration. Evaluations across a variety of image restoration tasks demonstrate that SP3 achieves perceptual quality comparable to state-of-the-art zero-shot diffusion and flow methods while being 3-630× faster.
Sean Man, Ron Raphaeli, Matan Kleiner +1
Technion – Israel Institute of Technology · Independent Researcher
We present the Large Processing Model (LPM), a diffusion-based generative framework for photorealistic video restoration under complex, in-the-wild degradations. To our knowledge, LPM is the first generative video restoration model deployed at industrial scale. LPM addresses the diverse degradations in user-generated content (UGC) through a unified system encompassing large-scale data engineering, foundation-model training, and efficient inference. Its enhanced architecture, progressive training strategy, and temporal-pyramid inference mechanism jointly enable high-fidelity, temporally consistent restoration of arbitrarily long videos across the broad content distribution encountered on UGC platforms. LPM has been deployed in production at Kuaishou, where videos processed by the model account for approximately 45% of total viewing time, delivering consistent improvements across key quality-of-experience metrics. Beyond perceptual enhancement, LPM delivers substantial system-level benefits: at comparable perceptual quality, it reduces bitrate by 20% relative to Kuaishou's in-house codec, yielding annual bandwidth cost savings on the order of hundreds of millions. Its low serving cost also enables integration into products such as Kling, demonstrating that generative restoration can be practical, scalable, and cost-effective for large-scale video processing.