Because semi-transparent is what makes it interesting.
95.55% is the closest it gets because in order to get 100% correspondence, you'd need enough base shapes to make every pixel uniquely colored. It's hard to estimate exactly how many shapes that would take because of the transparency, but with circles or any other very simple primitive, it's certainly far more than 250.
For some reason my comment disapeared. Wierd. I always wanted to try a GA that creates a program that draws an image, not the image itself. Probably using turtle graphics, i dunno.
Yeah. :) The different siblings will eventually run on different nodes/cpus, which will provide a 26× speedup. But it's still not particularly fast. I have some ideas for speed improvement, but since this is a background project (ha! — sorry) I've focused on the algorithm first.
It could be a nice way to store textures.
kristallize 2 months ago
Because semi-transparent is what makes it interesting.
95.55% is the closest it gets because in order to get 100% correspondence, you'd need enough base shapes to make every pixel uniquely colored. It's hard to estimate exactly how many shapes that would take because of the transparency, but with circles or any other very simple primitive, it's certainly far more than 250.
mattdmdotorg 2 years ago
For some reason my comment disapeared. Wierd. I always wanted to try a GA that creates a program that draws an image, not the image itself. Probably using turtle graphics, i dunno.
Houshalter 2 years ago
This looks great could we see some source code, Thanks!
Oktane999 2 years ago
Source code eventually. It's still kinda kludgy in places.
mattdmdotorg 2 years ago
hey! love the work man. anyway we can get a copy of this prog to try out ourselves?
bardcan 2 years ago
It says Elapsed time: 438 930s?
121 hours, 55 minutes and 30 seconds?
Is that correct?
fg1i 3 years ago
Yeah. :) The different siblings will eventually run on different nodes/cpus, which will provide a 26× speedup. But it's still not particularly fast. I have some ideas for speed improvement, but since this is a background project (ha! — sorry) I've focused on the algorithm first.
mattdmdotorg 3 years ago