I understood it less as transforming the images and more of deriving math formulas from the patterns in the image, closer to creating a bar graph to understand data than making a copy.