In what essentially could lead to a very cool Instagram filter, a team of German researchers have come up with an algorithm to convert any photograph into a prestigious work of art. Leon Gatys, Alexander Ecker, and Matthias Bethge from the University of Tuebingen have been exploring the unique capabilities of convolutional neural networks, which led to the discovery of the photo-turned-art project.
Based on the way in which complex neural networks process object recognition, the algorithm adapts photographs into stylistic works of art like those of Picasso, Vincent van Gogh, and other world-famous artists. Basically, the algorithm treats the style of art by specific artists as a texture, and filters the target image through several layers of computational units to create an image with features that are similar to the original works of art.
While it would make for an entertaining social media tool, the art project is just a byproduct of the group’s larger research, and demonstrates how the content and style of an image can be separated, even if these two elements do have to be carefully balanced to look anything like the original image.
Plus, the process actually takes an hour (still pretty impressive for generating images similar in style to the works of art that took the original artist’s lifetimes to create). While that timeframe could decrease as they optimize the process, you probably won’t actually see it coming to a social network anytime soon. But avid social media users can still dream, right?
Interested in reading more about these findings? You can read the original report, “A Neural Algorithm of Artistic Style,” online.
Read more Tech Tuesday features here.