I am writing an implementing of Karl Sim's technique for evolving pictures and textures with genetic programming (http://www.karlsims.com/papers/siggraph91.html), but I'm not sure what his "warped" image processing function do. In the paper he says: "The warped versions of functions take (U,V) coordinates as arguments instead of using global (X,Y) pixel coordinates, allowing the result to be distorted by an arbitrary inverse mapping function." The rest of the image processing operations are like img1 + img2 or sin(img), and describe color as a function of the X,Y pixel coordinates. How would the "warped" versions work with U,V coordinates?