We’ve read a lot about Google’s machine-learning projects over the past 12 months. Perhaps most intriguing was the Deepmind project which created works of art using neural networks. Or, perhaps a more accurate description is, that the DeepDream algorithm would turn existing pictures in to the stuff of nightmares. By distorting shapes in to animal heads and psychedelic patterns and colors, the finished product was almost terrifying.
As it turns out, Google put on an auction at a trendy San Francisco venue and sold some of its larger pieces for as much as $8,000.
As reported by WSJ, the whole process started with Google teaching its machine learning to recognize objects within images, to make searching for images more accurate and intuitive. Then, their engineers used that same technology to get the machines to transform the images, making them look more like those identified objects.
Google — along with a group of artists made up from engineers and actual artists — used a handful of different algorithms to create the unique works of art for the show. ‘DeepDream’ ran the aforementioned process multiple times to create weird, nightmarish images. ‘Fractal DeepDream’ was a variation on that process, while ‘Style Transfer’ essentially tries to create new images with the aim of copying a famous painting’s look.
As you’d expect, the auction at an old movie theater in the Mission neighborhood didn’t just attract the usual bunch of artsy San Fransiscans. Tech nerds and engineers likewise attended to hear just how a machine can create works of art. One of them was Amazon software engineer, Poorna Omprakash.
“I’m here because I want to know how they made the art,” said Ms. Omprakash. “But my second reaction is that it is beautiful!”
While traditionalists might bemoan the lack of physical paint or paintbrushes, I’m always up for seeing new mediums being used to create works of art. New technology allows us to create ‘paintings’ that weren’t possible before. Google’s Deepmind-powered pieces are prime examples of that. Still, I’m not sure I’d spend $8k for one, I think I’d rather just try to make my own.
Image Credit: WSJ