Just 2 weeks ago, Google’s “research blog” announced some interesting results from a program they had created to visualize what happens inside a neural network. When I say “interesting results”, I mean it gives scientists, artists, and even ordinary stiffs like me a possible peek into how a neural network — and maybe the human brain — internally processes data (like images and even sounds). And when you peel back the covers, so to speak, you can end up with results that look quite like what people see on psychedelics.
So a number of groups sprang up with a web interface to this open source software (my favorite is at DeepDreamr.com), and lots of people are finding it fascinating. Naturally, there’s even a subreddit for it now.
But I’ve also been hard at work putting a bunch of my own pictures through it, and after laughing at some of the results of faces with too many eyes and dog heads, I wondered what would happen with scenics. It turns out that quite a few of my recent pictures from NYC ended up looking really interesting! So, for those of you who’ve never seen NYC on LSD (and especially if you have!), these should prove very enlightening.
P.S. If you really like this stuff, it’s bigger and better on my Flickr album.