The sound files were composed using the following 2 algorithms, found in my algorithmic composer (not for sale still):
1) ReadRGB(image, scale, instrumentation, number of lines) Reads a line of image, gets RGB number of each pixel, gets RGB number modulo the length of the current scale, and
treats resulting number as a step number in current scale.
2)The idea is : the resulting score will be resembling an original image, if each instrument (in chosen instrumentation) has a fixed color. That is, the image was enlarged and the pixels, colored with colors that are not corresponding to color/instrument in current ensemble are colored white. This new image is treated as score with colored parts for each instrument.