Computer program reveals artists’ influences

Algorithm deciphers a painting's style by contents, composition, brushstroke

paintings by Frédéric Bazille and Norman Rockwell

A new computer program was able to detect a subtle link between paintings by Frédéric Bazille (left) and Norman Rockwell (right).

Wikimedia Commons; Shuffleton’s Barbershop © SEPS by Curtis licensing, Norman Rockwell Museum Digital Collection

Computers have become budding art historians. A new program can figure out a particular painting’s style — whether it’s abstract, impressionist or Baroque, for example — and tease out possible influences on the artists.

Program designers started with a close look at 1,710 paintings. For each work, the program logged the everyday items depicted, such as chairs and stoves, as well as features including color, composition and brushstroke. By comparing lists of these features with lists made from reference paintings, the program figured out the style of unknown paintings about 70 percent of the time, computer scientist Ahmed Elgammal and colleagues at Rutgers University in Piscataway, N.J., report August 19 in Multimedia Tools and Applications.

The program’s analyses could offer historians a new way to look at artists’ influences, Elgammal says. Among other links, the program proposed a connection between French impressionist Frédéric Bazille and American painting icon Norman Rockwell.

As the titles suggest, Diego Velázquez’s 1650 Portrait of Pope Innocent X (left) figured prominently in Francis Bacon’s 1953 Study After Velázquez’s Portrait of Pope Innocent X (right). Nathan Emory Coffin Collection of the Des Moines Art Center; Wikimedia Commons

Bazille’s 1870 Studio 9 Rue de la Condamine (at left in slideshow below) and Rockwell’s 1950 Shuffleton’s Barbershop (at right in slideshow below) look alike to the program. The two paintings have similar compositions (red lines) and objects (circled), along with a similarly positioned window (blue rectangles).

But the likeness may just be a coincidence, says Princeton art historian Emily Spratt. Two stylistically similar paintings don’t ensure the artists are historically related, she says. Still, she can imagine a program that looks at a painting and spits out information about the image and its place in history: “Maybe in the future, with computer vision technology, you could actually have a pocket art historian.”

A new computer program was able to detect a subtle link between paintings by Frédéric Bazille (left) and Norman Rockwell (right). (See next slide for highlights of similarities.) Wikimedia Commons; Shuffleton’s Barbershop © SEPS by Curtis licensing, Norman Rockwell Museum Digital Collection
Bazille’s 1870 Studio 9 Rue de la Condamine (left) and Rockwell’s 1950 Shuffleton’s Barbershop (right) have similar compositions (red lines) and objects (circled), along with a similarly positioned window (blue rectangles). Wikimedia Commons; Shuffleton’s Barbershop © SEPS by Curtis licensing, Norman Rockwell Museum Digital Collection

Meghan Rosen is a staff writer who reports on the life sciences for Science News. She earned a Ph.D. in biochemistry and molecular biology with an emphasis in biotechnology from the University of California, Davis, and later graduated from the science communication program at UC Santa Cruz.

More Stories from Science News on Computing

From the Nature Index

Paid Content