When the program is run it first chooses a random name. From the name it generates a mathematical formula which describes the color of each pixel in the picture. The program then uses the formula to generate the picture. The same name always determines the same picture.
A typical formula describing a pictures looks somewhat like this (it is the formula generated from the name random art):
The formula is given as an array of operators, which refer to other operators in the array. They comprise an acyclic directed graph. It is usually quite impossible to predict the appearance of a picture just by looking at the formula.
Instead of making up a random name, the program can also take a given name and generate a formula for it. You too may suggest a name and see what picture it makes in the visitor gallery.
It may happen that two different names determine the same picture, although this is rather unlikely due to the large number of all possible pictures. It is hard to determine their precise number, but we may safely claim that there are more random pictures than there are atoms in the universe.
The name of a picture normally consists of two words. The first word determines the colors and the layout of focal points in the picture, while the second word determines the selection and arrangement of graphical elements that appear in the picture. Therefore, if we change the first word and keep the second, the colors and the layout change while the style remains the same:
![]() foo gote | ![]() bar gote | ![]() baz gote |
In contrast, if we keep the first word and change the second, we obtain different pictures with similar selection of colors:
![]() random ballet | ![]() random art | ![]() random animal |
The rule about the first and the second word is only approximate, as in reality the second word also influences the colors and the first one the style.