 |
|
|
|
|
In 2001, David Hockney, the artist from Leeds, published his theory that great artists including Jan van Eyck and Caravaggio used lenses and simple cameras to 'trace' images onto canvas. But at the International Conference on Pattern Recognition (ICPR) in Cambridge on Thursday, a group of leading computer experts will show that a central image that he used to prove his theory shows clear signs of human error. Hockney said that the chandelier in van Eyck's celebrated 1434 painting known as The Arnolfini Portrait, was in such 'perfect perspective' that it must have been created with a concave mirror and pencil. In a book called Secret Knowledge: Rediscovering the Lost Techniques Of The Old Masters, and a BBC documentary, Hockney also theorised that world-famous painters such as Caravaggio and Bellini used the 'camera obscura' device. He said this pinhole technique, in which light passed through a small hole projects an upside-down image of the view on to an inner wall, was used to throw images of models on to a canvas where they would quickly be traced. But at the ICPR conference Dr Antonio Criminisi, an expert in computer vision analysis at Microsoft Research, will present new evidence that van Eyck's chandelier shows clear signs of error, and so is highly unlikely to have been created using optical aids. Criminisi has worked with Dr David Stork, visiting lecturer in art history at Stanford University in America, to commission a British realist artist to create another accurate chandelier image. This has been compared with that of van Eyck, other paintings and accurate photographs The images by Nicholas Williams, who is currently displaying work at St Mary's Cathedral in Edinburgh, were created solely by hand and eye and are more perfect than van Eyck's painting of the chandelier. Criminisi used computer technology to analyse the pictures, taking each arm of the chandelier and using a mathematical technique called homography to 'undo' the distortions of perspective and present a new front-on image without perspective. A perfect image would show each arm of the chandelier as exactly the same size; the percentage of error represents how accurately the image has been painted. Photographs showed an average degree of error of 3.1%. Chandeliers by Williams had a distortion of 8%, but the image with the most error, 12.6%, was The Arnolfini Wedding. This suggests that Hockney's theory is wrong, said Criminisi. The magnitude of error of the photographs always appeared in the bottom of results, showing that they were geometrically almost perfect. The images with greater error appear to have been constructed manually. Stork, said the computer techniques clearly disprove Hockney's theories Williams's achievement came from his doing what we know for sure artists of the early Renaissance did: practise careful drawing from life for years and decades. Williams's paintings are one piece in the emerging scholarly consensus Williams, whose experimental images will be on display for the first time at the Russell-Coates museum and gallery in Bournemouth later this year, said the work completely rebuts the theory that these painters used optical devices. It is important that this is resolved because it is misleading to younger artists looking at realistic paintings. Hockney, in looking for a Eureka moment, overlooked an alternative explanation that the artists simply worked it out for themselves. There is no question that optics have been used in art, but he is certainly wrong that this was as early as Caravaggio, and science has now proven him wrong. A spokeswoman at Thames & Hudson, which publishes Hockney's book, said
|
