When I first posted the “Evolution of Mona Lisa” post, there were a few questions raised regarding how well an evolutionary approach to image compression would work out.
A few days later, some guys at the NVidia forms started to try out the concept using hardware accellerated graphics.
That attempt was fairly successful, it was better than normal JPG, but worse than JPG2000.
Now there is another attempy made by Neil Graham, you can read his article about it here: http://www.screamingduck.com/Article.php?ArticleID=46&Show=ABCE
This attempt seems to deal much better with fine details than the CUDA version did.
I think that the final results shown in that article are awesome, although I would like to have seen some reference images using other compression algorithms with the same data size.