Eigenfaces is a very old technique for doing this (it was old when I implemented for the first time in 2000) and as you can see the results are rather blurry. It only takes about five lines of code in a matrix language -- collecting data was the main issue I recall.
If nothing else this serves as a good example of the value of framing your work. Done in a machine learning / stats context eigenfaces is not much more than a nice blog post (e.g. http://jeremykun.com/2011/07/27/eigenfaces/) or class project. But ... present it as an art project and you get frontpage on Google's new site.
Re:DevArt Art Made with Code
by m-photonic on Friday March 06, 2015 @02:02PM
(#9158807)
Very interesting. What if you took a large set of musical notes on various instruments, with about a two-second duration, with the same pitch, same volume, and same sample rate, and did the same thing? Eigensounds! And that's just with a time-domain representation of the sound -- you'd get different eigensounds if you used a frequency-domain representation.
I have to wonder if something like this is used by those song-recognizing apps that can tell you what a song is just by hearing a snippet through your phone.
Re:DevArt Art Made with Code
by MegaLeon on Friday March 06, 2015 @05:22AM
(#9156424)
Ah, Google DevArt. I partecipated last year and my project got a weekly feature award from the judges (every week they'd pick a promising project for that) but sadly didn't make in the final selection.
For some reason my project was pulled from the website after the competition ended, I tried to contact them about it but with no success. The project is on github, for those curious: https://github.com/MegaLeon/devart-arsmatrix
Re:DevArt Art Made with Code
by danboarder on Friday March 06, 2015 @02:45AM
(#9156033)
Is that because art is incomprehensible? But really the UX is not that bad -- there are filters across the top to reduce your view, restricting the results by filtering by Languages, Toolkits, Platforms, APIs, and Project Types. Seems reasonable.
For example, filtering by Toolkits>threejs shows some neat dataviz projects using threejs[1].
Re:DevArt Art Made with Code
by TheOtherHobbes on Friday March 06, 2015 @06:22AM
(#9156540)
Art is entirely comprehensible.
The problem with projects like this is that they often look more like technology studies than art.
There's a sweet spot between vintage media techniques and this modern(ist) everything-is-colour+code+concept digital style. But hardly anyone is exploring it.
Eigenfaces is a very old technique for doing this (it was old when I implemented for the first time in 2000) and as you can see the results are rather blurry. It only takes about five lines of code in a matrix language -- collecting data was the main issue I recall.
If nothing else this serves as a good example of the value of framing your work. Done in a machine learning / stats context eigenfaces is not much more than a nice blog post (e.g. http://jeremykun.com/2011/07/27/eigenfaces/) or class project. But ... present it as an art project and you get frontpage on Google's new site.
I have to wonder if something like this is used by those song-recognizing apps that can tell you what a song is just by hearing a snippet through your phone.
For some reason my project was pulled from the website after the competition ended, I tried to contact them about it but with no success. The project is on github, for those curious: https://github.com/MegaLeon/devart-arsmatrix
For example, filtering by Toolkits>threejs shows some neat dataviz projects using threejs[1].
[1] https://devart.withgoogle.com/#/?q=threejs
The problem with projects like this is that they often look more like technology studies than art.
There's a sweet spot between vintage media techniques and this modern(ist) everything-is-colour+code+concept digital style. But hardly anyone is exploring it.