This is a portrait of artist Hudson Marquez derived from a photograph provided by the artist. You can see photo in the gallery. cloudPainters algorithms used a melting pot of AI algorithms the most important of which were k-means clustering and feedback loops, to paint the portrait based on the progress it was making. At very beginning it took a photo of the canvas and compared it to Marquez's photo. It then calculated the areas with the biggest contrast and concentrates on reducing the contrast in that area. The heatmap in bottom right hand corner shows a visualization of this heatmap, and as the robot paints, you can see the image approaching white. It is basically trying to paint the differences between the canvas and the photo away. The one advance of this algorithm used in this painting for first time is that in past it would always start at white and only darken. Now based on a program I saw by Andrew Conru, it also lightens as needed. Blue means it should darken, while Red means it should lighten.