<p>Diffusion models have provided the state-of-the-art performances for the task of blind image super resolution thanks to their generative capability. The feature generation process of diffusion-based image super resolution schemes is performed over various temporal time steps. Despite this crucial factor, not much attention has been paid on developing a learning algorithm that by considering the temporal feature generation dynamic of diffusion process, be able to enhance the performance of the task of image super resolution. In this paper, we propose a novel learning algorithm for the diffusion-based image super resolution methods, which by employing the idea of operator learning, it is able to generate rich and representable sets of feature maps. In the proposed scheme, we model the dynamic of the diffusion-based image super resolution process over time using a differential equation, and then find its solution using the operator learning technique. The solution of the differential equation is then utilized to provide the guidance for the feature generation of the blind image super resolution network. The results of various experimentations have shown the superiority of the proposed scheme over the other state-of-the-art blind image super resolution networks in terms of various image quality assessment metrics.</p>