The article focuses on the emergence of an almost photographic realism, or ars nova, in paintings after 1425 at the dawn of the Renaissance. Contemporary artist David Hockney came up with a controversial theory, claiming that Renaissance paintings look realistic because artists used lenses and mirrors to project images onto canvases or similar surfaces and then trace and paint over the results. As part of an examination of this theory, other scholars and the author have used optical and computer-vision techniques to evaluate two of van Eyck's paintings that Hockney and his collaborator Charles Falco, a physicist at the University of Arizona, adduce as evidence. For a number of historical and technical reasons, Hockney envisions a camera obscura based not on a lens but on a concave mirror. Van Eyck's Portrait of Giovanni Arnolfini and His Wife (1434) is a key exhibit in Hockney's theory. To find the focal length of van Eyck's mirror, the author used a computer method developed by Antonio Criminisi of Microsoft Research in Cambridge, England, Martin Kemp of the University of Oxford and Sing-Bing Kang of Microsoft Research in Redmond, Wash. He was thereby able to adjust the radius of curvature of the mirror to "unwarp" the painted image. The depicted mirror, turned around, could not have been used as a projection mirror for the full painting. In fact, manufacturing a mirror from a blown-glass sphere that could have been used would have been beyond the capabilities of Renaissance technology. Other evidence throws doubts on the suggestion that van Eyck painted the Arnolfini portrait under a projection onto the oak panel. A projected image obeys the laws of perspective, but the perspective lines of the floor, window casement and other features in the painting do not meet at a vanishing point as they should. The perspective is consistently inconsistent. INSET: Overview/Analyzing van Eyck.