I will be collecting images generated by Evo-Lisa here and make a gallery out of it.
So if you have a portrait or other image that has been Evo-Lisa’ed that you want to show, please mail them to “Roger dot Alsing at Precio dot se”
For those of you who don’t know what I’m talking about, please see:
http://rogeralsing.com/2008/12/11/genetic-programming-mona-lisa-source-code-and-binaries/
And
http://rogeralsing.com/2008/12/07/genetic-programming-evolution-of-mona-lisa/
Art painted by evolution
I’ve already gotten the first image:
Mats Helander’s dog:
Bill Gates:
Moon man:
Scream:
Darwin:
Lara:
Tux:
Opera House Day time:
Opera House Night time:
Roger Johansson Blog 
Can we get statistics on those that are submitted with them (i.e. polygon count)?
sorry, that should have been “e.g.”
I have a butterfly that is an SVG conversion ( http://infinitefuture.blogspot.com/2008/12/image-decomposition.html ) and I wonder if it wouldn’t generate very quickly with a low poly. I saw another article at /. that referenced a person who created cartoons with CSS like this and could your algo create CSS ( Cascading Style Sheets ) or SVG ( Scalable Vector Graphics )?
From one programmer to another, that is simply amazing!!! I would LOVE to see the code and the suggestion above concerning stats. Awesome job. By the way, does your code take any image and automatically convert it to a polygonal, multilateral image?
Wow, that is a very effective algorithm. You should be proud.
I think you could speed up the code by changing 1 thing:
Apply the pixel comparison to the area covered by the polygon that has been changed only
what u think about it?
oh and for the geek within us,could you add a console under the original where we can see the modifications done in real time and what is happening like in this car evolution game,it’s pretty kewl and it would make a magnificent screensaver ^^
Car Evolution:
http://www.wreck.devisland.net/ga/
Roger,
Love that you’ve released the program. I’ll send my results over soon.
BTW I had an idea…I wonder what a scene from a film would look like if it was GV’ed (Generic Vectorized). That is, you take a scene from a movie, run GV on it until it gets to some desired fitness value, take the results, and play the GV images back.
Moreover, make this a distributed idea. Put up a bunch of frames from a video, have people run one image on their machine and send the results back to you. Distributed computing in action! Since each person’s result wouldn’t look quite the same (due to the evolutionary aspects of your computation) the resulting video may be quite freaky but cool to look at.
I’d do it on my laptop but it would take weeks, if not months, to render 1 minute of footage. But having 100 people work on it may make it feasible.
Regards,
Jason
Jason Bock, This can be done very easily with gimp when the images are created , it simple a matter of import directory as layers and save as GIF and it is an image that can be displayed without any special software compatibility issue. The gimp site has easy tutorials on how to do this and I also can build AVIs or any video format from a set of stills. Sounds like a neat idea. You could even use one of Google’s collaboration sites ( simultaneous multiple user access upload and download ( using a shared site key ) ) to have the images in a presentation and compiled from there by everybody. I looked into porting it to “C” with a make file for Linux and it is easily doable.
Paul,
I agree, it wouldn’t be hard for someone to do…it just takes a person to start organizing it (and I wish I had the time to do so!)
This is great. It’d be even better if it could export an SVG or generate a JPG from the result file… (seems like defeating the purpose in some ways I suppose ;))
Or is there some way to display the XML file outside of the program?
Roger, for my amusement I’m trying to replicate this without looking at your now-published code. I can get any old Mona Lisa image and use that, but I’d be interested in using the exact same one. I can’t find that anywhere except in screenshots, which are 256 color-gif images. Do you have the original around here somewhere? Thanks!
can I get a print of the Opera House at Night?
Chris, slightly higher-res Opera House shots linked from http://conceptdev.blogspot.com/2008/12/genetic-programming-in-art.html
Roger, since the images are vector-based, would it be possible to generate a printable version (ie. high enough resolution) if you could specify the width-height of a saved output image? Since you’ve published the code I will see if that is simple to do.
Great project!
Further to some of the comments above, I’ve written a little ‘Viewer’ app that opens .DNA files to save them. You can read about it here:
http://conceptdev.blogspot.com/2009/02/evolisa-viewer.html
It lets you save a high-res version
Or generate xaml to view in Silverlight
http://conceptdevelopment.net/wpf/EvoLisaViewer/silverlight.htm
I suppose SVG would be trivial from this point, if anyone really wanted it.
I find your Blog by Alain Lioret Blog.
I feel lots of energy in your work, Great!
you may be interested by looking at my Blog
http://bittleralain.wordpress.com/
sorry the first video is only in French.
Hey there, here’s the original Mona Lisa Pic after some 11 million generations (its been working for 3 days now).
Mona:
http://img219.yfrog.com/i/monac.png/
very interesting program, an upgrade will be necessary to do animation (input sequence of image). I would like to do one, when it become possible.
Alain
You should really try and make a high-res Darwin one for an art archive or something.
If you can get the processing time down you should package this as a plugin for Photoshop / GIMP.
very interesting, you could make a compression or encryption algorithm for images out of that maybe
hi guys , some tips for do something similar but in client side ?. HTML5 .. ?? … GWT ??..
I’ve done something similar in HTML5 – https://github.com/wolfgangmeyers/evolve.paint.digital
The tool is free to use and hosted at https://evolve.paint.digital