Generations of artists have found inspiration in New York City's vibrant, frenetic streets. From Alfred Stieglitz's photographs to Edward Hopper's realism to the ephemeral graffiti of Keith Haring—the city has been immortalized through countless mediums.
Now, a team of developers led by VFX artist Danil Krivoruchko has discovered a new way to envision NYC: through neural coding.
In a series titled "NYC FLOW," Krivoruchko created stylized impressions of the city by processing source video through an experimental neural network—an algorithm inspired by the learning techniques and recognition patterns within the human brain. According to the artist, the trippy-looking renderings were produced using an open-source project on Github that's available for anyone to play around with.
The code that Krivoruchko used is based on a paper by German researchers called "A Neural Algorithm of Artistic Style," which attempted to recreate the unique visual perception of humans in an artificial system. By developing an AI that could "separate and recombine content and style of arbitrary images, providing a neural algorithm for the creation of artistic images," the authors hoped to reveal how, exactly, humans perceive and create art.
Last year, Google's DeepDream program produced similar, albeit freakier, "artwork" using its own artificial neural network. By running images through a recognition system, DeepDream was able to spit out psychedelic, and highly exaggerated, impressions of whatever you fed it. Naturally, most people fed it porn.
And if you feel like taking a stab at this yourself, there's no need to go out and buy expensive video equipment. In a comment, Krivoruchko noted that his entire film was shot using an iPhone 6 in 240 frames per second. Since the footage ends up being heavily processed anyway, it doesn't really matter how fancy your gear is.
So now that you're feeling inspired, go forth, and make weird, nerdy art.