Pics / gifs / experimental

Basically you can take any shape you want to start with, and then scale it and rotate it down according to a set of functions that map points in a bigger domain to other points in a more restricted domain, which you then plug back in and repeat until your fixed points make an interesting pattern, which Falconer points out is usually a fractal. 

The reason I asked about IFS is because you can stretch, skew, and shrink shapes using affine transforms:

link to the code

(The first large part of the code (before “/////...///”) is simply something to render with motion blur.)

I don’t know much about IFS but this is implemented using recursivity. Each triangle has 3 children, and you start with one large triangle. To show one triangle you show its 3 children, excepted when you reached the maximum depth : in that case you just show one white triangle. The function “show” takes the positions of the 3 vertices of the triangle as input, and calls “show” on its 3 children (excepted when you reach max depth). The inputs for the new calls are computed with some interpolation between vertices with a random value (called “noise”) that loops through time.

If you have any question regarding the code, let me know.

Here’s an alternative version with other random values