Computer analysis of a 17th century painting shows that the artist did not, as has been claimed, use optical devices to project a perfect image of the scene onto his canvas. The researcher behind the analysis believes his findings undermine many aspects of a theory recently put forward by the painter David Hockney.
In his 2001 book Secret Knowledge, Hockney set out to show that the heightened realism of many Renaissance paintings was achieved by projecting images of the subject onto the canvas, which the artists then traced. This would have required artists to use a device such as a camera obscura.
But Hockney's theory is contentious among both art historians and physicists. It implies that from around 1420 artists were using sophisticated optics to project images onto the surfaces they were painting. Yet it was not until hundreds of years later, in the early 18th century, that artists like the Venetian Canaletto are generally acknowledged to have used such projectors. "The issues I raised have disturbed some people," Hockney says.
But next week, Stanford University physicist and art historian David Stork, who has been a fierce critic of Hockney's idea, will present evidence at the Electronic Imaging Conference in San Jose, California, US, that he believes show Hockney is wrong.
Stork has used computer imaging software to analyse the shadows in Georges de la Tour's 1645 painting Christ in the Carpenter's Studio in a bid to plot the direction and intensity of the light illuminating the scene.
This allowed him to determine whether the candle in Christ's hand was the only source of light. To illuminate the scene brightly enough to project it onto the canvas, de la Tour would have needed an external light source, probably the Sun.
Stork claims his analysis shows that a candle was indeed the only light source in the scene. He also says that given the type of lenses or concave mirrors available at the time, the brightness in the scene would have been reduced around 1000-fold at the canvas, making any projected image all but impossible to see and trace, unless several dozen oil lamps or hundreds of candles lit the scene.
As well as showing that the shadows cast can be plotted back to the candle, Stork's software indicates that the way light rays are reflected off Joseph's head are consistent with the candle being de la Tour's only light source.
This does not, however, convince Charles Falco, a physicist at the University of Arizona in Tucson, US, who worked with Hockney on his theory. "Artists would not have felt compelled to trace shadows as they were cast, but instead would have made the shadows consistent with the overall scene," he told New Scientist. But this would mean that the artist would have had to project this famous nighttime scene during the day and then completely rework the painting at night to make the light look realistic, Stork says.
Separate findings will be published in March by Thomas Ketelsen, a curator at the Museum of Prints, Drawings and Manuscripts in Dresden, Germany. Hockney has argued that the similarity between Jan van Eyck's drawing Portrait of NiccolÃ² Albergati and a larger oil painting of the same name could only have been achieved using optical projections.
But using a microscope, Ketelsen has found evidence of previously unseen pinpricks in the drawing - suggesting the copying method was mechanical, not optical. He suggests that a type of reducing compass called a "reductionzirkel" might have been used.
Falco points out that the pinpricks could have been made 50 years after van Eyck's death by someone wishing to copy it, or even 500 years after. "Holes can't be carbon dated," he says. But Stork thinks the mounting evidence can't be ignored. "The evidence doesn't support Hockney," he says.
"The debate is fascinating," Hockney says. "But it cannot end just because someone found pinpricks."