The neural network was taught to finish painting the paintings of the classics of painting
It looks like artists will soon be out of work. Artificial intelligence is mastering more and more creative professions: it no longer just writes poetry and draws abstract paintings, but also rules the works of immortal classics. The DALL-E 2 neural network was commissioned to finish painting the paintings known to everyone, and it coped so well that it seems that the author of the masterpiece himself worked.
The DALL-E 2 neural network was introduced in April 2022. She smartly generated images based on their text description, but soon everyone had enough of her. Later it turned out that we don't know about all the features of this system. The developers decided to check whether the neural network can finish painting other people's pictures. As it turned out, DALL-E 2 copes well with an unusual task and the augmented paintings look very realistic.
"Girl with peaches" by Valentina Serova received an augmented interior. Now she has a big window, shelves with all sorts of stuff and even a samovar.
"Morning in the Pine Forest" by Ivan Shishkin and Konstantin Savitsky received a wide viewing angle.
"Apotheosis of War" by Vasily Vereshchagin. Now we know what is left "behind the scenes". There is nothing particularly interesting there.
"Mona Lisa" by Leonardo da Vinci has legs now. It seems that this is much better.
Ilya Repin's "Boatmen on the Volga" got more foreground.
"Pink Mountains" by Nicholas Roerich. It turns out that a small village lurks at the foot of the mountains.
"Christ Pantocrator" from the Sinai Monastery. As and The Mona Lisa Savior got legs.
Ivan Aivazovsky's "Ninth Shaft" has become even more eerie due to the wide viewing angle and the bloody sunset.