Diffusion models have become the state-of-the-art generative model in a multitude of generative tasks such as audio and image synthesis. Until recently, there has not been much success with the specific image-to-image task ...