In November 2020, 25-year-old Russian programmer Ilya Belikov created the Neurussia community on the popular Russian social network VKontakte.
Once every few days, surreal images of Russian suburbs dotted with panel buildings and industrial zones appear on the group page. But this is not the work of an artist or photographer — they are all generated by a neural network.
“In Russia, ‘depressive culture’ in different manifestations is very popular. I got so fed up with the Runet [Russian internet] being littered with post-punk, panel buildings and doomers that I thought, if you can't beat ’em, join ’em — and become their leader.
And what could be more pointless and dreary than a neural network generating a hypertrophied chthonic [underworld] Russia,” Ilya Belikov explains the idea behind Neurussia in an interview with online news portal TJournal.
Belikov trained the neural network for three days by feeding it images of Russian panel buildings and suburban landscapes from other VK communities.
In total, the neural network analyzed about 15,000 photographs. Ilya says that it learned early on to copy the color of the faded apartment blocks.
“Like the view from my own window,” write Russian-speaking users, comparing the often depressing output of the neural network with the landscapes they see every day in real life.
As an added bonus, Ilya taught the neural network to generate images of Russian cats and platbands on wooden houses. Perhaps that will make Neurussia a jollier place? Hope springs eternal.