Shattered Sinet One
These images are generated by a model of a small network, inspired by ideas about neural networks in living creatures. The nodes of the network take numbers as their input, and transform them into other numbers which become inputs to other nodes in the network. The original numbers are derived from the x and y coordinates of the pixel, and the final numbers are represented as the red, green and blue colors for that pixel. The sine function is used as the nonlinear transfer function of the nodes, and so the name "sinet" is derived from the words "sine" and "network".
A process of artificial evolution inspired by Karl Sims' Genetic Images work from 1993 is carried out by a computer program, which varies the parameters of the sinet, measures certain characteristics of the output and keeps the variations that increase the measured value, which is to do with the complexity and diversity within the image. A final round of selection is carried out by the Artist (from a thousand candidates), and then the video is produced by varying some parameters of the sinet over the time span of the video.
The resulting works look liquid, glassy, flowing, refracting and reflecting.
In the Shattered Sinet series the Heaviside function is used as well as the sine function. This discontinuous function shatters the image into shards, making it look Cubist to the Artist.