The field of image restoration and generation is rapidly evolving, with a focus on developing innovative methods to improve image quality and diversity. Recent research has explored the use of diffusion models, vision-language models, and multi-task learning to address various challenges in image restoration, including degradation, noise, and blur. These approaches have shown promising results in enhancing image quality, reducing computational costs, and improving the robustness of image restoration models. Furthermore, researchers have investigated new guidance methods for diffusion models, such as adaptive guidance, token perturbation guidance, and feedback guidance, which have demonstrated significant improvements in image generation quality and diversity. Noteworthy papers in this area include UniRes, which proposes a universal image restoration framework for complex degradations, and SPARKE, which introduces a scalable prompt-aware diversity guidance method for diffusion models. Overall, the field is moving towards developing more efficient, effective, and flexible image restoration and generation methods that can be applied to a wide range of real-world applications.