Selective Hourglass Mapping for Universal Image Restoration Based on Diffusion Model proposes a novel method, DiffUIR, which combines strong condition guidance and shared distribution mapping for universal image restoration. The method is based on a conditional diffusion model and introduces a shared distribution term (SDT) to gradually map different distributions into a shared one. DiffUIR achieves state-of-the-art performance on five image restoration tasks, 22 benchmarks in the universal setting, and zero-shot generalization setting. The method outperforms existing universal methods with a model size of only 0.89M parameters, demonstrating its efficiency and effectiveness. The key contributions include the introduction of a selective hourglass mapping strategy, the validation of the distribution mapping strategy for universal image restoration, and the demonstration of the model's ability to meet real-world application demands. The method is evaluated on various image restoration tasks, including deraining, low-light enhancement, desnowing, dehazing, and deblurring, and shows superior performance compared to task-specific and universal methods. The experiments also show that the model can achieve zero-shot generalization on known and unknown tasks, further validating its effectiveness. The method is implemented using a diffusion model with a modified condition mechanism and SDT, and the results demonstrate its ability to capture shared information between different tasks, leading to improved performance in universal image restoration.Selective Hourglass Mapping for Universal Image Restoration Based on Diffusion Model proposes a novel method, DiffUIR, which combines strong condition guidance and shared distribution mapping for universal image restoration. The method is based on a conditional diffusion model and introduces a shared distribution term (SDT) to gradually map different distributions into a shared one. DiffUIR achieves state-of-the-art performance on five image restoration tasks, 22 benchmarks in the universal setting, and zero-shot generalization setting. The method outperforms existing universal methods with a model size of only 0.89M parameters, demonstrating its efficiency and effectiveness. The key contributions include the introduction of a selective hourglass mapping strategy, the validation of the distribution mapping strategy for universal image restoration, and the demonstration of the model's ability to meet real-world application demands. The method is evaluated on various image restoration tasks, including deraining, low-light enhancement, desnowing, dehazing, and deblurring, and shows superior performance compared to task-specific and universal methods. The experiments also show that the model can achieve zero-shot generalization on known and unknown tasks, further validating its effectiveness. The method is implemented using a diffusion model with a modified condition mechanism and SDT, and the results demonstrate its ability to capture shared information between different tasks, leading to improved performance in universal image restoration.