The neural network "corrected" the masterpieces of world art: how Monet and da Vinci were supposed to draw?
The AI model Stable Diffusion tried to reimagine the images of famous artists, and in some cases turned out to be very realistic.
The artist with the nickname Images_ai modernized the world's masterpieces of painting with the help of the Stable Diffusion neural network. She showed the results of the work on her Twitter page.
First of all, the user processed the "Mona Lisa" by Leonardo da Vinci. The neural network gave the woman in the painting sharper shapes and richer colors, and reworked the background.
"Honestly, da Vinci did it pretty sloppy by today's standards, but with the help of Stability Diffusion, we were able to bring it in line with modern trends," the blogger commented.
Johannes Vermeer's "Girl with a Pearl Earring" was criticized by the artist for the lack of details in her hair. The neural network "tweaked" the hairstyle in the portrait and made the character's face more realistic, thanks to which the picture looked like a photograph.
The blogger believes that the couple from the famous "American Gothic" by Grant Wood looks too pixelated by modern standards, besides, the clothes look pretty ridiculous. Stability Diffusion pulled out the faces of the characters, making them more caricatured, and re-dressed and replaced the village house with a modern urban backdrop.
“With the help of AI, we can see how Grant could rethink this work if he was still alive,” the artist said.
The Scream, the only painting by Edvard Munch that has become famous, consists predominantly of enigmatic swirls. The neural network reworked the background, assuming that it depicts a river and fjords, while the person in the picture stopped screaming.
Claude Monet was accused by a Twitter user of unprofessionalism. This French artist is considered the creator of impressionism in watercolor painting, a style characterized by vague and illegible forms.
“This is what the Water Lilies etude would look like if he really knew how to draw,” said Images_ai.
In the case of Sandro Botticelli's The Birth of Venus, the user did not like the environment, which she called "a bunch of blurry, strange and abstractly drawn figures." However, after processing with Stability Diffusion, objects became even more blurry and abstract, losing their original shape.