Comment on Genetic algorithms, Mona Lisa and JavaScript + CanvasComments−raverbashing13yThat's very interestingIt seems the default mutation parameters are a bit low, if you increase it it may lead to better resultsI think the main problem with this is that it's using a fixed number of polygons, and apparently is mutating their vertice positions.If it could be changed for a range of polygons (with a range of vertices, let's say from 3-6) maybe it can converge faster (and better)But it is nonetheless a very nice experiment. GA/GP is one of those things that really can work with "untractable" questions.
Comments
That's very interesting
It seems the default mutation parameters are a bit low, if you increase it it may lead to better results
I think the main problem with this is that it's using a fixed number of polygons, and apparently is mutating their vertice positions.
If it could be changed for a range of polygons (with a range of vertices, let's say from 3-6) maybe it can converge faster (and better)
But it is nonetheless a very nice experiment. GA/GP is one of those things that really can work with "untractable" questions.