What tool did you use to make the art on your website?

Your style relies heavily on recursion and similarities between scales, which conv nets are not good at, but programatic descriptions of images like LOGO are very good at. My strategy would be to manually write simple LOGO, python (or whatever tool you initially used) programs that generate each of the images on your site, and then prompt Chat-GPT with “write a program that generates an image combining ideas from <Program A> and <Program B>