Plug-and-play algorithms constitute a popular framework for solving inverse imaging problems that rely on the implicit definition of an image prior via a denoiser. These algorithms can leverage powerful pre-trained denoisers to solve a wide range of imaging tasks, circumventing the necessity to train models on a per-task basis. While denoising neural networks have traditionally served as restoration priors, recent works have proposed to incorporate more sophisticated models, such as super-resolution networks.In this presentation, I will show that any image restoration model can be used as an implicit prior for solving inverse problems. Our approach leverages that restoration models are trained to ensure a fixed-point property when composed with degradation operators, which yields an explicit prior formula. I will finish my presentation with a brief hands-on tutorial of the deepinverse library: https://github.com/deepinv/deepinv. If you are convinced to try it out, I’ll be happy to sit with you and help you with your implementations!