Now it's time for another tool. Here is it, my c# implementation of a genetic art tool based on sharpNEAT [1]. Inspired by the genart tool fom mattias fagerlund [2] which is based on delphiNeat[3] i create just for fun my own version of it with some new features like
- history (it's possible to go back in a previous generation and track another evolution)
- the user has more posibilities to affect the color calculations
- there are more options to influence the picture generating.
- math. expression (which will compile at runtime and affect the network signals)
- save/load several genomes
- save/load color schemes
take a look

here are some black/white generated pictures



and here some colorful



To run it you need .NET Framework 2.0 final. If you want play a little with it look here [4].
Or if you want to see the full source and extend it, download the whole Visual Studio 2005 solution [5].
Some tips:
- the more complex a network is, it's expedient to lightly increase the network iterations for better pictures.
- play with the mutation parameters, it's important for good results.
- play a little with the math. expression for the network input signals.
- tip: mirrorX/mirrorY in addition with "colored noise" and the "wired" color scheme gets the "real" pictures, then just play with the zoomfactor ;)
While i was coding this tool, a question is flashed in my mind. I see that all networks have the _same_ input signals, and these are constant. But all networks create an individual picture of the signal. Now, just map this to the human brains ... you see the parallelism? ... is all out there only an illusion?
life is a game.
[1] http://sharpneat.sourceforge.net/
[2] http://www.cambrianlabs.com/mattias/GeneticArt/
[3] http://www.cambrianlabs.com/mattias/DelphiNEAT/
[4] GenArt.zip, just the exe and some color schemes (130 kb), need .NET 2.0 to run
[5] GeneticArt full source (VS2005 Solution) (670kb)