random-art.org by Andrej Bauer generates individual images for user-inputted phrases. “The program accepts the name of a picture and uses it as a seed from which a picture is generated randomly. The same name always yields the same picture.”
It’s a fun, interactive artwork. The inclusion of the gallery where people can submit the images they’ve generated are really interesting because you can see all the different phrases people think of while they experiment with the generator. If you sort the gallery by popular, you can see the images voted ‘most interesting’ — one that I found most interesting was ‘sand dunes’, which by chance generated an image that looked like sand dunes.
I created a few myself:
Here is the artist’s explanation for how the code (Python) works:
The idea is to generate expression trees that describe an image. For each point `(x,y)` of the image we evaluate the expression and get a color. A color is represented as a triple `(r,g,b)` where the red, green, blue components are numbers between `-1` and `1`. In computer graphics it is more usual to use the range `[0,1]`, but since many operations are symmetric with respect to the origin it is more convenient to use the interval `[-1,1]`. I kept the program as simple as possible, and independent of any non-standard Python libraries.
If I changed anything about the project, I would like to see how each inputted word gets broken down and turned into an image.