Old Masters’ artworks could be exposed as having been painted by their understudies, thanks to new AI technology that reveals each artist’s brush stroke is as unique as a fingerprint.
Philip Mould, English art dealer and presenter of Fake or Fortune, said the technology signalled the “dawn of a revolution” that would allow academics to dissect a painting brush stroke by brush stroke.
Many of the Old Masters were often assisted by understudies, so the new technique would allow historians to establish just how much of their most renowned works the artists themselves were actually responsible for. It could also help to identify forgeries.
The technology has been described as potentially being as important to art as “DNA was to solving criminal cases”.
Mr Mould said that if the technology panned out in the way he hoped, it would “unquestionably” lead to “significant reassessments” of some of the great art collections across the world, including The Prado in Madrid, the National Gallery in the UK and the Met in New York.
The new AI technology – believed to be the first of its kind – analyses the 3D topography of a painting, revealing the slightest difference in elevation in the contours of the paint.
The machine learning technology was able to correctly identify, with 95 per cent accuracy, the correct artist from tiny brushstrokes roughly the size of a single paint brush bristle.
Artist comparison: 3d topography
Kenneth Singer, professor of physics at Case Western Reserve University in Cleveland, Ohio and the lead researcher, said the technology "stunned us by actually being able to tell with very high accuracy whose hand among four different art students had painted even a tiny fraction of a brushstroke".
“We’ve uncovered what could be considered the unintentional style of a painter.
“There is something unique to each person, perhaps a physiological element, related to how the paint leaves the brush.”
How does the technology work?
Prof Michael Hinczewski, co-researcher, said this level of precision essentially helps to develop a “fingerprint” for each artist.
“We were shocked with what emerged,” he said. “This is something that must somehow be intrinsic to each artist, not something that the painter is consciously trying to do.”
Elizabeth Bolman, professor of art history at the university, worked alongside the scientists. She said the new technique would “add a level of data to art attribution that humans can’t discern”.
The team’s research was recently published in the journal Heritage Science.
Many of the most notable masters, from El Greco to Rembrandt to Rubens, employed workshops that included other artists who painted parts of the canvases, to speed up production and meet market demands for their art.
Leonardo Da Vinci may only have painted part of The Annunciation, right
Credit: AP Photo/Andrew Medichini
Mr Mould, who has been an art dealer since his teenage years, said he had been waiting for a breakthrough like this for over a decade.
He said: “If this pans out as we hope, it could do it would be analogous to the introduction of DNA in the solving of crimes.
“As a tool to understand art, it is transformative. We could be at the dawn of a revolution in attribution.”
While Mr Mould acknowledged that the technology was very much in its infancy, he said its impact on the art world could be “radical”.
The attribution of Sandro Botticelli's 15th-century painting, Young Man Holding a Roundel, has been disputed by several analysts
Credit: AP Photo/Seth Wenig
He believes paintings by artists such as Joshua Reynolds, Thomas Gainsborough, Rembrandt and Rubens could potentially be affected by the new techniques.
He said that the next step would be to “amass as much data and as much information as possible and codify it”, so that it could be used alongside established historical knowledge.
History of disputed paintings
“It could have a very considerable impact on the categorisation of artists on all sorts of levels,” he said.
“It could lead to some very significant reassessments in collections across the world.”