AI Used to Reproduce ‘Lost’ Picasso Nude

A painting that Pablo Picasso once painted over himself has been rediscovered thanks to some advanced 3-D printing and artificial intelligence (AI) techniques. George Cann and Anthony Bourached, who are currently studying for their doctorates in machine learning at University College London, put together the innovative method for revealing the image of a nude beneath what Picasso had painted over. In 1903, the Spanish-born artist used the canvas he’d previously painted the nude on to produce another image of a person in an artwork that became known as The Blind Man’s Meal. This meant his previous painting had remained hidden for over a century, until now.

The newly rendered version of the nude figure has been given the name The Lonesome Crouching Nude by the company that uncovered it. Oxia Palus, the technology firm set up by Cann and Bourached, used the latest digital techniques to resurrect the lost artwork. In fact, superimposed X-rays had been used in the past to produce a fluorescent version of the image before, so it had been known it was there for some time. What Oxia Palus has been able to do is to bring the hidden artwork back to life in a much more realistic way.

Developing Technologies

Cann and Bourached used the previously utilised x-ray-based imaging technique to come up with an outline of the hidden painting beneath the one that viewers can actually see. They then programmed artificial intelligence algorithms to add brushstrokes in the style of Picasso’s hand to the x-rays. The programming meant that Oxia Palus could generate a map of the height of the paint as it was applied to the canvas originally. Unlike conventional methods, this meant a greater level of texture and granularity could be achieved with the ensuing digitised image. Once the nude portrait had been captured by the software system, the next step was for the PhD students to print what had been gleaned from the picture onto a new canvas using 3-D printers.

Bourached said that when it is broken down in this way, any piece of art can be seen as a store of complex information. That might be too much for individual people to take on board but with machine learning helping to analyse such information, it is now possible to handle all of the complexity involved. According to Bourached, this will mean that people can now understand their history and culture better. The postgraduate student admitted that x-ray images were useful in art history and restoration projects but that AI adds another substantial layer to the mix.

According to Cann, the image the pair have revealed would not have been one that Picasso would have wanted to paint over if he could have avoided it. He pointed out that the original artwork dates back to the artist’s so-called Blue Period when, early in his career, he may not have had enough money for new blank canvases. “At the time that he produced [these images]… Picasso was poor,” Cann explained. “Artistic materials were expensive back then, so he probably painted over the first image with a degree of reluctance,” he added.

New Perspectives

In a statement issued by the pair, Cann said that he hoped Picasso would have been happy knowing that this hidden treasure was finally being revealed for future generations to enjoy. He noted that this achievement came some 48 years after the artist’s death and well over a century since the painting was originally concealed.

David Dibosa, of the Chelsea College of Arts in London, said that everyone who valued Picasso’s work would be energised by the emergence of a new image by him. Dibosa, who runs a Masters course in curating and collections, reckoned that the combination of technology at play was particularly impressive. However, he questioned why the portrait had to be reprinted on canvas and posed the interesting idea that, perhaps, the work could have been left in the digital realm for people to view with a device.