A conceptual illustration of computational art history.

‘Computational Art History’ Finds Clues in the Canvas

Stories You May Like

After 50+ Years, Computer Graphics Pioneer Remains a Powerhouse Prof

The Swim Test: A (Wet) Rite of Passage for Undergrads

The ABCs of C-A-Ts

An emeritus professor taps his engineering acumen to explore the materials used by Van Gogh, Vermeer, and more

By Beth Saulnier

Lisa Pincus calls it “the ugly Vermeer.” Titled Young Woman Seated at a Virginal, it depicts a rosy-cheeked lady with light brown curls, wearing a yellow shawl and gazing vaguely in the direction of the viewer as she fingers the keys of the title instrument, a type of harpsichord. The painting, an oil on canvas currently held by the Leiden Collection in New York, is believed to date from the early 1670s, a few years before Johannes Vermeer’s death in 1675.

But to Pincus—a Cornell art history professor who specializes in 17th-century Dutch art—it’s arguably the weakest entry in the artist’s distinguished oeuvre. “I think it’s a really wooden depiction,” she says. “We have a pretty set idea, ‘This is what Vermeer does; this is how his paintings look.’ It doesn’t have the subtlety, the nuance, the kind of light we expect of Vermeer.”

As Pincus explains, there’s a long history of Vermeer forgeries. And for years, the authenticity of Young Woman Seated at a Virginal was in doubt, given its perceived artistic shortcomings. “I still would like to disown it,” Pincus says. “The only reason I don’t is because Rick has made it clear that it is by Vermeer.”

“Rick” is C. Richard Johnson Jr., an emeritus professor of engineering—and, at the risk of a mixed metaphor, something of a Renaissance man. Johnson, who joined the Cornell faculty in 1981, has spent decades teaching and doing research in electrical engineering, particularly in the fields of control systems and signal processing. But over the past 14 years, his interests have entailed as much art as science.

Professor Rick Johnson seated in his office in front of three versions of a painting of Van Gogh's bedroom at Arles.
Professor Emeritus Rick Johnson in his office in Rhodes Hall, where the décor includes framed posters of Van Gogh's classic painting Bedroom in Arles as seen via different types of imaging techniques. (Photo by Robert Barker)

A pioneer in the field of computational art history, Johnson leverages both his engineering acumen and his abiding passion for art to study the physical materials with which works are made. In addition to paintings on canvas, Johnson has turned his computational eye on other media such as photographic paper, parchment, and the silk used in ancient Chinese paintings; most prominently, he and colleagues have done extensive analysis of the paper used in centuries-old works, including Rembrandt prints and the codices (scientific notebooks) of Leonardo da Vinci.

The power of counting threads

In 2007, at the start of a five-year stint as an adjunct research fellow at the Van Gogh Museum, Johnson launched the Thread Count Automation Project (TCAP), which informs the study of works on canvas by mapping the thread density of the fabric on which they’re painted. By comparing the “weave maps” of various paintings, researchers can establish that they came from the same roll of cloth—and, therefore, that the artworks were likely produced in the studio of the same artist around the same time.

That’s how Pincus was convinced of the authenticity of Young Woman Seated at a Virginal, whose weave map matched that of one of the most beloved of Vermeer’s works: The Lacemaker, housed in the Louvre in Paris.

In a December 2018 lecture in the A.D. White House as part of the Milstein Program in Technology and Humanity, Johnson put up a slide showing the two paintings: Young Woman was on the left, The Lacemaker on the right, and their respective (and visibly matching) weave maps beneath them.

“If you doubted the painting on the left, how could they be on the same canvas?” he asked the audience. “Somebody’s answer would be, ‘His cousin painted it; he showed him how and gave him the canvas; it’s not a Vermeer.’ So [the match] doesn’t cure everything, but it’s another piece of forensic evidence that stacks up to make the case.”

Images of two paintings by Vermeer, "Young Woman Seated at a Virginal" and "The Lacemaker," are seen with their matching weave maps.

Young Woman Seated at a Virginal (top left) and The Lacemaker (top right), each oriented to correspond with their respective weave maps (bottom row). (Image provided)

In other words—as Johnson frequently stresses—weave matches are a powerful tool, but they don’t themselves constitute proof of authorship; that has to come through additional means, such as understanding of an artist’s technique and studio practices. For Pincus, it’s that very knowledge that makes this match probative.

“We know very little about Vermeer, but we’re pretty sure he didn’t have studio-mates, and he didn’t have students,” she says. “He wasn’t rich. He died bankrupt. I don’t think he’d be cutting off pieces of canvas [and giving them away]. I’ve reconciled myself that this is a late edition to the corpus.”

'No slouch'

As Johnson Museum of Art curator Andy Weislogel, PhD ’00, puts it, Rick Johnson is “no slouch”; his art-and-science bona fides are impeccable. Until retiring with emeritus status in July 2021, he was Cornell’s Geoffrey S.M. Hedrick Senior Professor of Engineering. (While he and his wife have sold their Ithaca-area home and moved to Hawaii, he has retained his office in Rhodes Hall and is continuing to oversee grant-funded research for the next two years.)

On the other side of the Atlantic, Johnson has a research appointment at the Rijksmuseum in Amsterdam—while he wasn’t able to travel there due to the COVID pandemic, he aims to return—and has served as an adviser to the Netherlands Institute for Art History in The Hague, where he was the first-ever fellow in computational art history.

Johnson’s many admirers include Cornell’s own Frank Robinson, director emeritus of the Johnson Museum and himself an expert on 17th-century Dutch prints. “A work of art is the meeting place of many different disciplines; that’s what’s so interesting about studying this stuff,” observes Robinson, who invited Johnson to present his research to museum staff during his time as director. “The scientific aspect of it is absolutely vital—and in that respect, Rick is top of the line.”

Stories You May Like

After 50+ Years, Computer Graphics Pioneer Remains a Powerhouse Prof

The Swim Test: A (Wet) Rite of Passage for Undergrads

“A work of art is the meeting place of many different disciplines; that’s what’s so interesting about studying this stuff. The scientific aspect of it is absolutely vital—and in that respect, Rick is top of the line.

Frank Robinson, director emeritus of the Johnson Museum

But, as Johnson cheerfully admits, he’d never even been to an art museum until he was in his early 20s, as an undergrad studying abroad in Germany. At the end of the term, he spent two months traveling around Europe, and the many museums he visited instilled a lifelong love for the Dutch masters. As a doctoral student in electrical engineering at Stanford, he talked his way into a graduate seminar on Rembrandt and went on to earn a PhD minor in art history—the first the university had ever granted.

A focus on Van Gogh

In the mid-2000s, Johnson resolved to integrate his engineering research with his passion for art; he landed a meeting at the Van Gogh Museum, where he ultimately spent a semester. He initially got involved with work to identify fakes through x-ray analysis of brushstrokes and other features—the subject of a May 2007 interview he did on NPR’s “Science Friday”—but ultimately focused on thread counting as a method with the potential for broad applications.

As Johnson explains, the practice then entailed the use of magnifying glasses and the painstaking, manual counting of small sections of x-radiographs. Not only was it tedious, but counting an entire canvas—rather than a number of sample sections—was wildly impractical. “Let’s say you have a painting that’s 45 by 50 centimeters, and you’re going to count the threads in every little half-centimeter,” he says. “You end up with something like 9,000 places to count. Nobody’s going to do it.”

But Johnson immediately realized that counting threads was an example of a basic engineering concept: frequency, or the number of times that a particular thing occurs within a set interval. He attacked the problem by writing a basic algorithm called a Fourier transform—something Cornell engineering students study sophomore year—which is essentially an equation for reverse-engineering information into its constituent parts.

“It worked, and I took it back the next day and said, ‘We’ll count every painting in the museum!’” he recalls with a chuckle. “And they said, ‘Please, try not to be so American.’”

Validation for Sunflowers

In the intervening years, Johnson and his colleagues have refined their computational methods and gone on to map much of Van Gogh’s oeuvre, matching his works to nearly four dozen individual rolls of canvas. Together with existing documentation—such as the artist’s frequent correspondence with his brother Theo—the weave maps offer valuable insights into Van Gogh’s career, which lasted a mere ten years.

“We can actually put the rolls back together, pretty much,” Johnson says, “and that will date his paintings to within a three-week period.”

Six Van Gogh paintings and their corresponding weave maps.
Weave maps (seen in the two bottom rows) connect three sets of paintings by Van Gogh, showing that each pair came from the same roll of canvas. (Image provided)

Among their revelations: that the seventh in Van Gogh’s Sunflowers series—bought for nearly $40 million in 1987 by a Japanese insurance magnate despite being long considered a potential forgery—was in fact likely genuine, given that it matched other works on a twenty-meter roll of jute canvas supplied by fellow artist Paul Gauguin.

“It didn’t look like Van Gogh’s brushwork, but it was a coarse fabric, and he was struggling with it,” Johnson explains, noting that the artist complained about the difficulty of painting on jute in a note to Theo. “[The previous experts] didn’t take that into account.”

And Johnson’s weave-mapping work has gone beyond Van Gogh and Vermeer. At the talk in the A.D. White House, he put up slides of three of Claude Monet’s famed Haystacks paintings—and punctured the enduring romantic notion that the artist returned to the same farm field throughout the year to capture the seasonal nuances.

“He did them all in his studio,” he said. “They’re all from the same roll of canvas.” The method also helped art historians identify the subject of a Velázquez portrait—a dwarf in the Spanish royal court—by demonstrating that the work was painted around the same time as the artist’s portrait of King Philip IV in the mid-1640s.

A true Poussin?

In one of the talk’s most intriguing anecdotes, Johnson described the analysis of three seventeenth-century French paintings: The Triumph of Pan, The Triumph of Bacchus, and The Triumph of Silenus. While it’s known that Cardinal Richelieu commissioned the series from artist Nicolas Poussin for his chateau, only the first (in London’s National Gallery) has always been considered a true Poussin. As Johnson and colleagues explained in the Journal of the American Institute for Conservation in May 2013: “The success of this commission led to demand for the production of copies close in date to the originals, some of which are represented in major collections.”

For that reason, the authenticity of Bacchus (housed in Kansas City’s Nelson-Atkins Museum of Art) has been questioned over the years, though it’s now generally accepted as original and the museum labels it as such. And as for Silenus? As Johnson told the audience back in late 2018: “It’s in the basement in the National Gallery in London and is considered to be a fake [from] 100 years later, and nobody needs to see it”—but as weave mapping revealed, all three came from the same roll of canvas.

The august museum has since come to agree with him. As it now states in its online description of Silenus: although the painting’s uneven finish and other factors had led some scholars to consider it to be a copy, “recent conservation treatment and technical analysis point instead to this being Poussin’s original.”

Top image: Photo illustration by Cornell University.

Published October 5, 2021


  1. Frank Robinson

    What a wonderful article! A truly fascinating discussion of a fascinating scholar, who has made an important contribution to art history.
    Best regards,
    Frank Robinson

    • VE, Class of 2017

      I agree, Frank. Great article on a very interesting project! I remember some years ago taking the elevator with you in the Johnson after we’d attended a talk. You turned to me and said (as I was discussing my research) “Remember, a good dissertation is a done dissertation!” I appreciated your sage words that afternoon and often mulled them over when I felt myself stalling. I hope you are well! Cheers from the PNW.

  2. Diane Matyas

    I appreciate the information about how an artist works, experiments, fails, achieves, and manages their work- all revealed by these tech methods. Artists laboring in their studios all know this- but so often scholars want to have a “perfect” pedigree, as if they paint and draw like machines. They are just humans in love with battling on to make pictures.

Leave a Comment

Once your comment is approved, your email address will not be published. Required fields are marked *

Other stories You may like