The labeled co-registered image is segmented (zoomed in on Madonna) and subsequently a set of augmented segments are produced that vary in brightness, contrast, and sharpness. These augmented segments are then passed to the trained Leonardeschi model and inference is run, transforming the augmented segments in a set of paintings. After an exploration of the set of paintings (possible solutions) the most aesthetically pleasing result is manually selected. The final painting, Madonna, shows a resurrected Leonardo da Vinci.