Well after a bit of hitting myself over the head with quite how simple it was I’ve managed to generate some hyperbolic Julia sets:

The problem I was having is that I was getting hung up over the distance function rather than letting the algebra take care of it for me. The big advantage of our approach is that one can just ‘twist a knob’ at the start and the method works the same in Euclidean geometry as Hyperbolic. I was trying to re-scale my point twice for no good reason effectively collapsing me down to the origin or flying off to the Poincaré disc rim at each iteration.
Both pictures show the set from the same starting point (I think) although they differ quite a bit because (I imagine) the hyperbolic Mandelbrot differs. In the hyperbolic case I have mapped the Poincaré disc back onto the plane (i.e. the rim has been pushed to inifinty) to more clearly show the structure of the fractal although one can plot it on the disc. The colours are assigned using the normal ‘minimal iterations to test for inclusion’ principle (I have some shortcuts for the axes which is why there are some artifacts there).
It looks like the hyperbolic set is far more ’swirly’ than their Euclidean counterparts with a fine internal structure. Should be fun playing with the 3d version now — although I need to re-cast a load of complex analysis to deduce a suitable distance function.
Update: There now exists a high resolution version showing the internal stucture nicely. This one is on the Poincaré disc so the blue part around the outside should be considered not part of the space.
Update 2: A hyperbolic Mandelbrot has also been generated. If the Euclidean one looks like a squashed bug I think this one looks like a roast turkey.


