Composite image degradations, where rain, haze, low-light, and snow overlap spatially within a single frame, pose a fundamental challenge for unified restoration models. We identify two failure modes on such inputs. First, deterministic flow transport averages over feasible solutions and collapses to over-smoothed predictions. Second, image-level expert routing yields a selectivity ratio of only 1.2 × , with near-uniform weights that leave experts under-specialized on localized distortions. Motivated by these observations, we propose F 2 D-Net, a factorized stochastic transport framework. The model factorizes the restoration velocity field into a shared backbone and degradation-specific expert increments. A learned spatial gating then assembles them at every pixel. The core mechanism is a state-dependent multiplicative noise whose magnitude scales with the local residual. The noise remains large far from the clean target to encourage diverse reconstructions, and contracts to zero at convergence to preserve fine detail. The closed-form log-normal transition collapses sampling to one network call, matching feed-forward speed. On CDD-11, F 2 D-Net attains the best structural and perceptual quality across all complexity tiers with substantially higher expert selectivity than image-level routing. It also remains competitive on standard three-task and five-task all-in-one benchmarks. Our project page is at https://sxvvv.github.io/f2net .

Ultra-high-definition (UHD) image restoration is vital for applications demanding exceptional visual fidelity, yet existing methods often face a trade-off between restoration quality and efficiency, limiting their practical deployment. In this paper, we propose TSFormer, an all-in-one framework that integrates Trusted learning with Sparsification to boost both generalization capability and computational efficiency in UHD image restoration. The key to sparsification is that only a small amount of token movement is allowed within the model. To efficiently filter tokens, we use Min- $p$ with random matrix theory to quantify the uncertainty of tokens (lower trustworthiness), thereby improving the robustness of the model. Our model can run a 4K ( $3840\times 2160$ ) image in real time (40fps) with 3.38 M parameters. Extensive experiments demonstrate that TSFormer achieves state-of-the-art restoration quality while enhancing generalization and reducing computational demands. In addition, our token filtering method can be applied to other image restoration models to effectively accelerate inference and maintain performance.

In the realm of image super-resolution, learning-based methods have made significant progress. However, limited computational resources still restrict their application. This prompts us to develop an efficient method for achieving effective image super-resolution. In this letter, we propose a novel adaptive feature selection modulation network (AFSMNet) tailored for efficient image super-resolution. Specifically, we design feature modulation blocks, which include the adaptive feature selection modulation (AFSM) module and the self-gating feed-forward network (SFN). The AFSM module dynamically computes the importance of each feature channel. For channels with differing levels of importance, we employ distinct processing strategies, thereby concentrating the computational resources of the network on the more critical features as much as possible. This approach facilitates the maintenance of a low computational cost without compromising performance. The SFN restricts the flow of irrelevant feature information within the network through a simple gating mechanism. In this way, our method achieves efficient and effective image super-resolution. Extensive experiment results show that the proposed method achieves a better trade-off between reconstruction performance and computational efficiency compared to the current state-of-the-art lightweight super-resolution methods.