Madonna of the Rose, Raphael (Raffaello Sanzio). Credit: Prado National Museum
A famous painting by Raphael features a face that was NOT created by the Renaissance master, according to a new artificial intelligence analysis.
Experts from the universities of Nottingham, Bradford and Stanford used in-depth intelligence-based trait analysis on the Madonna della Rosa (Madonna of the Rose), and they discovered that although most of the painting is indeed by Raphael , Joseph’s face is probably not from the same hand.
Della Rosa’s work, on display at the Prado Museum in Madrid, Spain, has intrigued art experts, with some, notably Raphael specialist Professor Jurg Meyer zur Capellen, suggesting that Raphael’s student Giulio Romano might have contributed to it. One theory is that the rose and the lower part could have been painted by someone else. However, new analysis reveals that the lower part of the painting is “most likely” by Raphael.
Professor Hassan Ugail, director of the Center for Visual Computing and Intelligent Systems at the University of Bradford, designed the algorithm that recognizes authentic works by Raphael, with 98% accuracy.
Professor Ugail explains: “When you present an image to the computer, it will give a binary classification of whether or not it is an authentic Raphael, with 98% accuracy. We can now say with great confidence whether a painting is an authentic Raphael or not. .
“When we tested the della Rosa as a whole, the results were inconclusive. So we tested the individual parts and, although the rest of the image was confirmed to be that of Raphael, the face of Joseph appeared to be probably not Raphael’s.”
“This analysis of artwork by artist Raphael presents an objective and quantifiable approach, using machine learning, to the classification of painted images. It promises to be a useful additional tool in future investigations of this nature, alongside well-established methods such as spectroscopy. It is adaptable in that the works of other artists can be examined using the same technique, and this is the aim of future research,” said Dr. Christopher Brooke, honorary research fellow at the University of Nottingham.
A paper on the work, titled ‘Deep Transfer Learning for Visual Analysis and Attribution of Paintings by Raphael’, co-authored by Professor Ugail, Professor Brooke, Professor Howell Edwards (Emeritus Professor of Molecular Spectroscopy, University of Bradford) and Stanford University Assistant Professor David G. Stork, was published in the Heritage Sciences newspaper.
Howell Edwards, Emeritus Professor of Molecular Spectroscopy at the University of Bradford, explains more of the history of the Madonna della Rosa: “Painted on canvas around 1517/18, the Madonna della Rosa was considered by early connoisseurs to be an autograph of Raphael . that is to say, he painted 100% of it. Beginning in the mid-1800s, art historian Johann David Passavant and others questioned its composition and preferred to attribute its execution partly to Raphael and partly to his workshop.
“The attribution to Raphael’s workshop was gradually accepted later and attributed in particular to his student Giulio Romano and perhaps also to Gianfrancesco Penni. In Spain, the original attribution has never been questioned.
“Some connoisseurs consider that the quality of the composition and painting of the Virgin, Child and St. John far exceeds that of St. Joseph, which they believe was added to the workshop as an afterthought.
“The analysis of our work by the AI program has conclusively demonstrated that while the three figures of the Madonna, the Child Jesus and Saint John the Baptist are unequivocally painted by Raphael, that of Saint Joseph is not is not and was painted by someone else – perhaps by Romano, as zur Capellen and others believe.
The research team has already used artificial intelligence-assisted computer facial recognition on a mysterious painting known as de Brécy Tondo, which resembles Raphael’s Sistine Madonna. The computer determined that it was a work by Raphael, based on an earlier pigment analysis carried out by Professor Howell Edwards, also at the University of Bradford, which placed it firmly in the Renaissance era.
“Through extensive feature analysis, we used images of authenticated paintings by Raphael to train the computer to recognize his style in great detail, from brushstrokes, color palette, shadows and every aspect “The computer sees much deeper than the human eye, at the microscopic level,” said Professor Hassan Ugail, director of the Center for Visual Computing and Intelligent Systems at the University of Bradford.
Professor Ugail continues: “This is not about AI taking people’s jobs. The process of authenticating a work of art involves examining many aspects, from its provenance, pigments, condition, and more. can be used as a tool to facilitate the process.
David G. Stork, an assistant professor at Stanford University who pioneered the application of computer vision to problems in the history and interpretation of fine art paintings and drawings, also contributed to the recent research. He shares Professor Ugail’s opinion that this type of analysis is a tool in the process of authenticating a work of art, to be used in addition to traditional methods.
Dr Stork, author of Pixels & painting: Foundations of Computer-Assisted Connoisseurship, said: “Computer methods are slowly but surely proving that they can aid traditional humanistic studies of art, but they must always be employed with a in-depth understanding of the historical context of art. , and their results understood and interpreted in the broader context of artistic knowledge relevant to the problem at hand.
“The attribution and authentication of works of art are among the most arduous and difficult tasks faced by art specialists who must study provenance (the document of ownership, sales and exhibitions of a work), the materials (chemistry of pigments, canvas, paper, varnish), the condition of the work (is it degraded over time or retouched), the iconography (the people and objects represented are appropriate) and finally the know-how (in-depth visual study of brushstrokes, color, composition, etc.).
“Most computational studies in the field of art, including the current study, have focused on improving the know-how of connoisseurs. The results of the current study should not be considered sufficient to an authentication decision, but as a step toward improving overall authentication protocols. Some of the most successful computational studies of art have mined vast databases of art images to learn style of an artist and other properties.
“As such, databases grow, computer algorithms are refined, and, most importantly, as humanist art scholars critique and refine computational methods, computational methods will improve and become widely used in art history and criticism.
More information:
Hassan Ugail et al, Deep transfer learning for visual analysis and attribution of Raphael paintings, Heritage Sciences (2023). DOI: 10.1186/s40494-023-01094-0
Provided by the University of Nottingham
Quote: AI study shows Raphael’s painting was not entirely the master’s work (December 21, 2023) retrieved December 21, 2023 from
This document is subject to copyright. Apart from fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for information only.