You can’t actually move into a Roy Lichtenstein painting, but you’re getting close. A new photo-editing app called Prisma was recently released (to no small amount of fanfare) with a simple goal: rechristening even your most ill advised selfies into works of art. The app is available for free on IOS and Android.
Prisma is very simple:
1) Select (and crop to square) the photo you want to use, and
2) Select one of the app’s preset ‘filters’
The app then connects to its cloud servers and combines the two. While this all sounds deceptively simple, the device actually connects to a neural network, which draws an entirely new image based on the pair of images the user has selected. These neural networks, although undertaking the seemingly simple task of mashing two images together, actually ‘draw’ an entirely new image, allowing the user to toggle how much weight the second image is given.
The ‘filters’, as it were, are actually a very diverse group of visual references and styles. While they do include expected nods, such as to the above-mentioned Lichtenstein’s pop art, Alphonse Mucha’s Art Nouveau portraiture and the Great Wave, they also include some unexpected references, such as to Roland Palmaerts, a Belgian watercolorist, Natalie Ratkovski, a German Illustrator and Francis Picabia, a French painter who’s work is associated with cubism, impressionism and dada. (Van Gogh’s Starry Night is conspicuously absent.) It’s a clever way to reinterpret images, as well as an innovative crash-course in the dos-and-don’ts of basic composition.
The technology behind computer generated art
While much of this app’s popularity is based on its viral novelty, the use of neural networks to perform human-like tasks, including but entirely unlimited to art, has been the subject of intense academic focus for sometime.
Artificial neural networks, such as the code Prisma uses, are heavily tied to what is expected to be the creation of artificial intelligence and have been instrumental in producing the programming and technology that makes modern life possible. Auto-piloting programs, voice recognition and even e-mail spam filtering are all made possibly through the use of this technology.
The Russian social site Ostagram.ru, which lets users combine any two photographs of their choosing, predates Prisma, which was also developed in Russia. Ostagram.ru was in turn influenced by Google’s Deep Dream project. This was another project ostensibly about combining images, but was actually an experiment in teaching an artificial neural network how to produce images in pre-existing styles.
Google made the code for Deep Dream open source in last year, and various apps and experiments have popped up since then, Prisma being the most popular.
Pros and Cons
Prisma’s success is helped by the fact that user interface is incredibly simple. The general layout is similar to Instagram’s editing interface, but more streamlined. Whereas Instagram allows uses to tilt and shift their pictures obsessively without any regard to filters, Prisma merely allows users some minimal cropping before transforming their images.
As it is with most photo-altering apps, the various transformations are displayed along the bottom of the device. During peak hours, the pictures may take a bit of time to render, but, ultimately, are worth the wait.
It is, however, unclear if the artists who created the images, or their estates, were compensated for the use of their images. This is especially troubling, given that at least some of the images used are fan art or tribute paintings. Nothing on the app or on their website provides this information. While it is certainly possible that the artists were compensated, it is not wholly unlikely they weren’t.
Overall, however, this free app is enchanting and well worth checking out.