So I'm still reading through the code in an attempt to understand how all of this fits together, but how do you think blue noise (or my recent attempts at approximating it) would affect the rendering? https://observablehq.com/@jobleonard/pseudo-blue-noise https://observablehq.com/@jobleonard/pachinko-tree ... actually, I think I can make my algo return a list that meets the requirement of plist. BRB, forking this notebook and doing some hands-on testing :)

https://observablehq.com/d/30b40039d28802dd#split1D Ah, nvm, I didn't realize the distribution of the plist doesn't affect the distribution of the output in a different way

I don't know so much about blue noise, but plist is a list of constants that defines a distribution - i.e. how to pick the functions in the IFS. To introduce noise, I guess one would need to fiddle more directly with the function that does the selection. That function could depend on n, the number of iterates, for example; or even on the iterates themselves.

I'm not sure if you noticed, but I've got a whole collection of notebooks built on top of this one. A major objective that I've been working towards is the description of the boundary between adjacent fractal tiles - basically a generalization of the ideas in this notebook: https://observablehq.com/@mcmcclur/the-twindragon-with-boundary

Thanks for the explanation! Yeah, I came back here from one of the other ones you recently published actually! :)