The simulation of the raindrop refraction is not intended to be a faithful representation of physical reality, but one that produces visual results that please me. So, the amplitude of the wave train does not decay in the same ways as a real wave, and I also allow "twisting" refraction effects (deflection of rays at right angles to the radial direction) which could only be produced in unusual optical materials (and definitely not isotropic fluids).
I good deal of experimentation goes into each of these images, looking for the right balance between complex detail and overall coherence. Firstly one needs to choose the right type of original photograph, generally one with some simple, strong features and a relatively limited and complementary colour palette. Then I adjust various details of the underlying mathematical algorithm to produce just the effect that I am seeking. Although there is a lot of trial and error, it is very much guided by past experience in what is likely to move the image in the way I desire, and remains very much like creating art with conventional media, in that I try to go in one direction and sometimes get surprised by unexpected and delightful results.
