Image-to-Image covers AI techniques that take one image as input and generate a new image, often incorporating style transfer, variation models, and refinement algorithms.