I’ve had a creeping sense of dread ever since i saw this article about Robbie Barrat‘s computer-generate art over the weekend.
Barrat, a recent high school graduate in West Virginia, made the images by feeding thousands of classical nude paintings scraped from WikiArt into a Generative Adversarial Network (GAN). The GAN uses a system of two neural networks called a “generator” and a “discriminator” to create convincing versions of the works using data from the paintings and machine learning.
When he previously tested this technique with landscape oil paintings, Barrat says the GAN was able to produce fairly convincing compositions with some surreal accents. In the nude portrait experiment, however, the neural network refused to move past its Dalí period.
“The GAN didn’t successfully learn how to make realistic nude portraits,” the 18-year-old Barrat told me via email. “The discriminator part of the GAN isn’t really able to tell the difference between blobs of flesh and humans, and once the generator realized it could keep feeding the discriminator blobs of flesh, and fool it this way, both networks just stopped learning how to paint more realistically.”
Usually the machine just paints people as blobs of flesh with tendrils and limbs randomly growing out – I think it’s really surreal. I wonder if that’s how machines see us…
Barrat posted many of the final pieces of artwork — which can only be described as surreal, blobby, swirly naked women — on Twitter. It’s almost like a very intoxicated Salvador Dali and a dizzy Picasso joined forces to make art. …Barrat’s AI-assisted artwork isn’t exactly sensual. In fact, most of the nudes look like they are melting on a very hot day.
“The way that it paints faces makes me uncomfortable. It always paints them as like, purple and yellow globs — that isn’t in the training set so I’m actually still not sure why it does that.
Check out the deeply unsettling results below. (via boingboing)
Dreaming about the latent space of flesh pic.twitter.com/xWCS5Op5ls
— Robbie Barrat (@DrBeef_) April 1, 2018