Clarification on the de Brécy tondo that could be by Raphael according to artificial intelligence

A singular piece of news has been circulating in recent hours: an artificial intelligence believes that a tondo derived from the Sistine Madonna could be attributed to Raphael, based on its resemblance to the original. No wonder it looks like him: it's a copy!

A singular piece of news concerning the Tondo de Brécy, an ancient copy of Raphael’s Sistine Madonna, has bounced around in Italian and international newspapers in recent days: two British universities, Nottingham University and Bradford University, have in fact usedartificial intelligence to compare the tondo with its model, going so far as to determine, as the BBC (the first to spread the news) writes, that facial recognition technology has discovered that the painting “is most likely a masterpiece by Raphael.” The BBC reports that the similarity between the Madonnas in the two paintings “was found to be 97 percent, while a comparison of the Child in both works produced a similarity of 86 percent.”

There is no shortage of statements from researchers who took advantage of artificial intelligence to compare the two paintings. “From the direct facial comparison,” Dr. Christopher Brooke, an honorary researcher at the University of Nottingham, an expert in digital image analysis and co-author of the paper on the alleged discovery (not yet published), told the BBC, “it turns out with a 97 percent match, a very high statistical probability that the artworks are by identical creators.” Further confirmation, Brooke continued, would come “from analysis of the pigments used in the Tondo, which showed that the characteristics of the painting are considered typical of Renaissance practice and therefore highly unlikely to be a later copy.”

Still, Hassan Ugail, professor of visual computing at the University of Bradford, said that “Looking at faces with the human eye shows an obvious similarity, but the computer can see much deeper than we can, in thousands of dimensions, at the pixel level. Based on the high evaluation of this analysis, along with previous research, my fellow co-authors and I concluded that identical models were used for both paintings and are undoubtedly by the same artist.”

The work, held in the private collection of collector George Lester Winward (the De Brécy Trust Collection, named after his French ancestors) who acquired it in 1981, has often been studied in the past, and has never been convincingly attributed to Raphael. The most authoritative opinion to date is that of Jürg Meyer zur Capellen, who has ruled out any attribution to Raphael or his school. “The manner of painting,” he wrote in an expertise, “is not comparable to Raphael’s original works, such as the Sistine Madonna itself. Both paintings are painted on canvas, and the beautiful modeling of the Sistine Madonna clearly shows the structure of the underlying canvas, as do other paintings on canvas, such as the Veiled Madonna (Palatine Gallery, Florence) and the Portrait of Baldassarre Castiglione (Louvre). These are two of the few examples of the use of canvas as an image support in Raphael’s work. Compared to these examples, the pictorial layer of the Tondo de Brécy is absolutely compact and does not allow any conclusions about the image support. Obviously the priming and painting produce a rather hard layer that covers the canvas underneath. The result is craquelé on the surface of the tondo, which is absolutely foreign to all Raphael paintings. Alongside this, the X-ray shows a manipulation of color not found in the artist’s autograph paintings.” Technical analyses performed on the painting in the early 2000s found that all the pigments used by the artist who executed the Tondo de Brécy predate the 17th century, but of course the fact that it is an ancient work does not necessarily imply its attribution to Raphael.

The analyses also found some pentimenti, judged, however, to be “not relevant” by Meyer zur Capellen, who “does not rule out the idea that copies were made in the early 17th century of the Sistine Madonna on the basis of the original cartoon.” Basically, the scholar concludes in his expert opinion, “an attribution to a specific painter must be made on stylistic analysis. So a comparison between the tondo and the Dresden Sistine Madonna shows that the latter offers more differentiation of details, such as hair, limbs and rendering of clothing, etc. The modeling of the Child’s body is very soft and has subtle shades of light in shadow, a quality not achieved in the tondo. The coloring of the latter is characterized by a slight greenish tone, which stands in contrast to the warm colors of the Dresden Sistine Madonna. Many other differences could be found. From this point of view Raphael or his school in my opinion cannot be the authors of the tondo. This is not to say that the tondo does not have a good pictorial quality, but it was done by another author and in another time.”

As for the results of artificial intelligence, it is quite obvious that the computer finds a strong resemblance between the Tondo de Brécy and the Sistine Madonna, since the former is a copy of the latter and the copyists tended to try to be as faithful to the original as possible. “It certainly does not require artificial intelligence,” art historian Fabrizio Federici wrote on the Mo(n)stre page, “to say that the faces of the figures look quite similar, a copy tries to look like the original, otherwise it would not be a copy.” We are also talking about a painting, the Sistine Madonna, which had an enormous fortune for a long time, from the time it was painted to even the present day (think of the two little angels in the lower portion of the painting). And numerous copies of the work exist, some of excellent quality. The qualitative gap between the Sistine Madonna and the Tondo de Brécy, properly identified by Meyer zur Capellen, is precisely what prevents an attribution of the latter to Raphael, despite the fact that artificial intelligence seems to suggest otherwise. And then, artificial intelligence also identifies a percentage gap between the two paintings, but the news is very steamy, and even the University of Bradford website, on which the news of this research came out, says nothing more than what the BBC reports: in short, it is not clear whether the computer analysis is purely quantitative, which would make it lose all credibility, or it is also qualitative, and therefore it also compared modeling, drawing, glazing, transparencies, color range differences, and so on.

And even the market considers it unlikely to be a work by Raphael. In 2013, what the Financial Times called “a bizarre attempt to sell online what the seller believes is a prototype for part of Raphael’s famous Sistine Madonna” was staged. The work was put up for sale in a unique self-administered online auction, with a base of £1 million. It failed to attract a single bid.

Pictured: left detail of the Sistine Madonna, right detail of the Tondo de Brécy.

Clarification on the de Brécy tondo that could be by Raphael according to artificial intelligence
Clarification on the de Brécy tondo that could be by Raphael according to artificial intelligence

Warning: the translation into English of the original Italian article was created using automatic tools. We undertake to review all articles, but we do not guarantee the total absence of inaccuracies in the translation due to the program. You can find the original by clicking on the ITA button. If you find any mistake,please contact us.