Researchers at University College London have used artificial intelligence to reimagine a piece of art that Picasso painted over with “The Old Guitarist” (1903-04). The hidden painting features a nude woman sitting with one arm outstretched.
On the canvas, signs of her presence are faintly visible in the texture of the paint. Her eyes gaze out above the guitarist from behind the grey-blue wall of paint. But in 1998, conservators at the Art Institute of Chicago used X-rays and infrared light to create a radiograph that shows both “The Old Guitarist” and the lost woman overlapping. That provided proof of her existence and revealed she had been painted in her entirety before Picasso reused the canvas. The painting also matches a sketch of a woman that Picasso sent to a friend.
After researchers manually edited the combined image to remove the guitarist, they could apply artificial intelligence to the black-and-white figure that was left. Using a neural network capable of overlaying artistic styles to other images, researchers Anthony Bourached and George Cann transferred the style of Picasso’s “La Vie,” a pre-1904 blue period painting, to the outline of the lost woman. The result was the same figure, now white with pale blue shadows, sitting in front of a dark blue backdrop with black ripples.
The technology they used was first published in 2015, and its creators provided a preview of its abilities by applying the stylings of Vincent van Gogh’s “Starry Night” to a clip of the 2002 animated movie “Ice Age.”
Picasso was known to reuse canvases, painting over artworks that museums would be proud to display today. Bourached and Cann also looked behind Picasso’s “The Crouching Beggar” and reimagined a mountainous landscape that art historians believe was painted by Santiago Rusiñol. And in 2014, a portrait of a man was identified behind Picasso’s “The Blue Room.”
It’s impossible to recreate a painted-over image. But digital versions can reimagine the lost works in the style of artwork created by the same artist or at the same time in their artistic careers. Even a recreation of a lost piece of art can be useful to art historians and conservators.
The authors say in their paper, “Our method of combining original but hidden artwork, subjective human input, and neural style transfer helps to broaden an insight into an artist’s creative process.”