Can anti-terrorist technology help identify lost sitters?
May 8 2012
Probably not, as I conclude in today's Independent:
Software developed to recognise terrorist faces is being adapted to solve the mystery of portraits of unidentified people.
In certain cases, cutting-edge "face recognition" technology could identify faces from digital images, detecting similarities in facial constructs. The data will come from scans of known features of individuals, such as in a death mask or identified sculpture.
A feasibility study is being conducted by two art historians and an electronic engineer at the University of California. They describe FACES (Faces, Art and Computerised Evaluation Systems) as a "new tool for art historians". The project has received a $25,000 government grant.
Conrad Rudolph, professor of art history at the university, said: "Before the advent of photography, portraits were, almost by definition, depictions of people who were important in their own worlds. But, as a walk through almost any major museum will show, a large number of these unidentified portraits from before the 19th century have lost the identities of their subjects."
Frans Hals' The Laughing Cavalier, the 1624 masterpiece in the Wallace Collection, London, is among famous portraits whose sitters remain unknown. The picture's title was coined in the 19th century. Jeremy Warren, the Wallace's director of collections, said: "With the Laughing Cavalier, everyone accepts that name, but actually he's not laughing and he's not a cavalier ... I'd love to know who he is. If this technology can help us do it, we'd be absolutely delighted."
Bendor Grosvenor, a specialist in portraits at the Philip Mould Gallery, London, would particularly like to identify a "rather beautiful portrait" by an anonymous 17th-century hand – currently on show at the National Portrait Gallery.
He said: "It was traditionally called The Duke of Monmouth on his Deathbed, but it isn't him as the dates don't work. Deathbed portraits are relatively rare, so who was important enough, or loved enough, to have been painted in such a moving portrayal by a good artist? I would love to know."
But he added: "Most unknown sitters are unknown because they were only painted once, and there is no other likeness with which to compare them. So the new programme will most likely only help with portraits of people for whom we already have other portraits."
Professor Rudolph accepts that "difficulties are inherent" through variations in expressions, age, angle of pose and lighting. But initial tests – on identified 15th-century portraits of Lorenzo de' Medici, the Florentine ruler – have shown how faces can be reduced to labelled graphs and matched up.
This idea has been floating around for a while now. The late Labour MP and former sports minister Tony Banks (for whom I used to work, and who was a great collector of historical portraits), suggested using hooligan-spotting cameras to look for lost portraits as far back as the '90s.
I suspect that such technology has a long way to go before it can be really useful to art history. Portrait painters invariably had their own way of drawing faces, which can make comparing likenesses in works by other artists incredibly difficult. Sir Peter Lely is a well known example - as Pepys said of his portraits, 'good, but not like'. So the same sitter in a Lely can look quite different when painted by Kneller. Probably, this new computer programme will never be as useful as a well-studied art historian with a good memory for faces.
One day, computers may well be able to not only identify sitters, but artists too. Then I'll be out of a job. But I reckon we picture hunters and art historians have a few years left in us yet.
Update - a reader writes:
Regarding that deathbed portrait, whoever the man was there does seem something slightly ominous about his hidden neck. Was he, perhaps, someone who had his head cut off and later sewn back on again? Is that why it was thought to have been the Duke of Monmouth? [...] it seems to me that after death one wouldn't need to be wrapped up against the cold anymore, at any rate.
Update II - another reader writes:
Thinking about your excellent comment on the quondam Duke of Monmouth NPG, I looked up beheaded persons, of whom they are happily very few. I wondered about Sir Henry Vane the Younger and Colonel Penruddock, who both seem vaguely similar in face and age and date of death. The tight lips drawn up lips of the NPG sitter intrigued me, so I began to google 'postmortem effects of decapitation' but then I remembered that's the sort of thing loonies look up.