Can AI Give Us a Real Look at Lost Masterpieces?

In 1945, fire claimed three of Gustav Klimt’s most controversial paintings. Ordered in 1894 for the University of Vienna, “the Faculty Paintings” —as it is known — is unlike the former work of the Austrian symbolist. When he presented them, critics were baffled by their remarkable separation from the aesthetics of the time. They were immediately rejected by university professors, and Klimt left the project. Shortly thereafter, the works were found in other collections. During World War II, they were placed in a castle north of Vienna for storage, but the castle burned down, and the paintings probably accompanied it. What is left now are some black-and-white photographs and writings from the period. Still, I focused on them.

Well, not the paintings themselves. Franz Smola, a Klimt expert, and Emil Wallner, a machine learning researcher, spent six months combining their expertise to revive Klimt’s lost work. It’s a laborious process, one that starts with black-and-white photographs and then incorporates artificial intelligence and a lot of intelligence about the artist’s art, to try to recreate what the lost looks like. drawings. The results are what Smola and Wallner showed me — and even they were shocked by the captivating technicolor images produced by the AI.

Let’s make one thing clear: No one is saying that this AI goes back to Klimt’s original works. “It’s not a process of re-creating the actual colors, it’s the re-coloring of the photos,” Smola quickly noted. “The medium of the photograph is already an abstraction from real works.” What machine learning does is provide a glimpse into something that is believed to have been lost for decades.

Smola and Wallner find this appealing, but not everyone supports AI filling these gaps. The idea of ​​machine learning to recreate lost or damaged works, like Faculty Paintings itself, is controversial. “My main concern is about the ethical dimension of using machine learning within the context of conservation,” says art conservator Ben Fino-Radin, “because of the many ethical and moral issues. issue with beat the field of machine learning. “

To be sure, the use of technology to revive works of human art is fraught with thorny questions. Even if there is a perfect AI to know what colors or brushstrokes Klimt uses, there is no algorithm that can create an authorized purpose. Debates about this have been going on for centuries. In 1936, before Klimt’s paintings were destroyed, writer Walter Benjamin argued against mechanical copying, even of photographs, saying that “even the most perfect copying of a work of art is lacking. of an element: its presence in time and space, its uniqueness. its existence in the place where it takes place. ” This, Benjamin wrote The Work of Art in the Period of Mechanical Reproduction, is what he calls an act of “aura. ” For many art lovers, the idea of ​​a computer reproducing an intangible element is absurd, if not absolutely impossible.

And yet, there is much more to learn from what AI can do. Faculty Paintings was instrumental in Klimt’s development as an artist, an important bridge between his more traditional first paintings and later, more radical works. But what they look like in full color remains shrouded in mystery. That’s the puzzle Smola and Wellner tried to solve. Some projects, organized by Google Arts and Culture, not about perfect procreation; it’s about giving a glimpse of what’s missing.

To do this, Wallner developed and trained a three -part algorithm. First, the algorithm feeds hundreds of thousands of art images from the Google Arts and Culture database. It helps to understand objects, works of art, and composition. Next, it is specifically studied in Klimt’s drawings. “It creates a bias in his colors and motifs over time,” Wallner explains. And finally, the AI ​​is fed with color cues on specific parts of the drawings. But with no color references in the drawings, where do these signs come from? Even Klimt expert Smola was surprised by the many details revealed in the writings of the time. Because the paintings are considered more ugly and bizarre, critics are likely to describe them at length, down to the artist’s color choices, he said. “You could call this a historical irony,” said Simon Rein, the project’s program manager. “The fact that the drawings caused a scandal and were discarded puts us in a better position to restore them because there is a lot of documentation. And those kinds of data points, if fed by the algorithm, create a more accurate version of how these drawings might have looked at the time.

The key to that accuracy lies in matching Smola’s skill algorithm. His research reveals that Klimt’s work at this time was likely to have strong standards and consistency. The study of existing paintings from before and after Faculty Paintings gives clues to the colors and motifs that recurred in his work at that time. Even the surprises that Smola and Wallner encountered are confirmed by historical evidence. When Klimt first appeared in his paintings, critics noticed his use of a red that was, at the time, unique in the artist’s palette. but The Three Ages of Woman, painted shortly after Faculty Paintings, boldly using red, a Smola believes the same color caused the commotion first seen in Faculty Paintings. The writings from the period also elevate a color and cry about the shocking green sky in another Painting Faculty. Matching these writings with Smola’s knowledge of Klimt’s particular palette of vegetables, when fed by the algorithm, is what makes one of the first shocking images from AI.

Source link


Leave a Reply

Your email address will not be published. Required fields are marked *