Talk abstracts

When computers look at art:
Image analysis in humanistic studies of the visual arts


David G. Stork

www.diatrope.com/stork/FAQs.html

New computer methods have been used to shed light on a number of recent controversies in the study of art. For example, computer fractal analysis has been used in authentication studies of paintings attributed to Jackson Pollock recently discovered by Alex Matter. Computer wavelet analysis has been used for attribution of the contributors in Perugino's Holy Family. An international group of computer and image scientists is studying the brushstrokes in paintings by van Gogh for detecting forgeries. Sophisticated computer analysis of perspective, shading, color and form has shed light on David Hockney's bold claim that as early as 1420, Renaissance artists employed optical devices such as concave mirrors to project images onto their canvases.

ArnolfiniCoverGoodSmall

How do these computer methods work? What can computers reveal about images that even the best-trained connoisseurs, art historians and artist cannot? How much more powerful and revealing will these methods become? In short, how is computer image analysis changing our understanding of art?

This profusely illustrate lecture for non-scientists will include works by Jackson Pollock, Vincent van Gogh, Jan van Eyck, Hans Memling, Lorenzo Lotto, and others. You may never see paintings the same way again.

Joint work with Antonio Criminisi, Andrey DelPozo, David Donoho, Marco Duarte, Yasuo Furuichi, Micah Kimo Johnson, Dave Kale, Ashutosh Kulkarni, Gabor Nagy, M. Dirk Robinson, Silvio Savarese, Morteza Shahram, Ron Spronk and Christopher W. Tyler

Did early Renaissance masters 'cheat' by tracing optical images?
Computer image analysis and art history address a bold theory


David G. Stork

www.diatrope.com/stork/FAQs.html

In 2001, artist David Hockney and scientist Charles Falco stunned the art world with a theory that, if correct, would profoundly alter our view of the development of image making. They claimed that as early as 1430, some Renaissance artists secretly employed optical devices such as concave mirrors to project images onto their canvases, which they then traced or painted over. In this way, the theory attempts to explain the newfound heightened naturalism or "opticality" of painters such as Jan van Eyck, Robert Campin, Hans Holbein the Younger, and several others. Before scholars had a chance to examine the theory, it received a great deal of public promotion.

Theories are not judged by the fame or fervor of their proponents or un-peer-reviewed assertions, of course, but rather by the consensus of large number of independent experts examining all evidence pro and con, presented and evaluated through scholarly peer review or special conferences devoted to evaluating the theory.

This talk for general audiences, profusely illustrated with Renaissance paintings, will present the results of the first independent examinations of the Hockney tracing theory, including the work of 15 scientists/technologists, 8 historians of optics or art, three curators and two award-winning realist artists. It covers peer-reviewed data, dramatic physical discoveries, new computer image analysis techniques, and a review of 15th-century technology and the documentary record with special emphasis on Lotto's Husband and wife (1543), van Eyck's Portrait of Arnolfini and his wife (1434), Caravaggio's The calling of St. Matthew (1599-1600), Campin's Mérode Altarpiece (1425) Memling's Flower still-life and others. The results of these scholars in different disciplines, all lead to a remarkably consistent—indeed unanimous—conclusion about the tracing theory. You may never see Renaissance paintings the same way again.

Joint work with Antonio Criminisi, Andrey DelPozo, David Donoho, Marco Duarte, Yasuo Furuichi, Micah Kimo Johnson, Dave Kale, Alexander J. Kossolapov, Ashutosh Kulkarni, Gabor Nagy, M. Dirk Robinson, Silvio Savarese, Morteza Shahram, Ron Spronk, and Christopher W. Tyler.

Computer vision in the study of art:
New rigorous approaches to the study of paintings and drawings


David G. Stork

www.diatrope.com/stork/FAQs.html

New rigorous computer algorithms have been used to shed light on a number of recent controversies in the study of art. For example, illumination estimation and shape-from-shading methods developed for robot vision and digital photograph forensics can reveal the accuracy and the working methods of masters such as Jan van Eyck and Caravaggio. Computer box-counting methods for estimating fractal dimension have been used in authentication studies of paintings attributed to Jackson Pollock. Computer wavelet analysis has been used for attribution of the contributors in Perugino's Holy Family and works of Vincent van Gogh. Computer methods can dewarp the images depicted in convex mirrors depicted in famous paintings such as Jan van Eyck's Arnolfini portrait to reveal new views into artists' studios and shed light on their working methods. New principled, rigorous methods for estimating perspective transformations outperform traditional and ad hoc methods and yield new insights into the working methods of Renaissance masters. Sophisticated computer graphics recreations of tableaus allow us to explore "what if" scenarios, and reveal the lighting and working methods of masters such as Caravaggio.

ArnolfiniCoverGoodSmall

How do these computer methods work? What can computers reveal about images that even the best-trained connoisseurs, art historians and artist cannot? How much more powerful and revealing will these methods become? In short, how is the "hard humanities" field of computer image analysis of art changing our understanding of paintings and drawings?

This profusely illustrate lecture for scholars interested in computer vision, pattern recognition and image analysis will include works by Jackson Pollock, Vincent van Gogh, Jan van Eyck, Hans Memling, Lorenzo Lotto, and several others. You may never see paintings the same way again.

Joint work with Antonio Criminisi, Andrey DelPozo, David Donoho, Marco Duarte, Micah Kimo Johnson, Dave Kale, Ashutosh Kulkarni, M. Dirk Robinson, Silvio Savarese, Morteza Shahram, Ron Spronk, Christopher W. Tyler, Yasuo Furuichi and Gabor Nagy

Computer vision and the new connoisseurship:
Bringing science to the science of close readings of art

David G. Stork

www.diatrope.com/stork/FAQs.html

Within the history of academic art history and scholarship, the subdiscipline of connoisseurship—the close technical and formal reading of works—fell out of favor in the 1970s.  The tradition of Giovanni Morelli (1816–91), Bernard Berenson (1865–1959) and their followers for authentication, identification of hands in a work, tracing of stylistic influences among artists, and establishing esthetic benchmarks (about "quality") was largely rejected as something only curators should concern themselves with, not academic scholars.  Connoisseurship was parodied as an elitist endeavor wherein a monocled expert made subjective judgments in a refined, sometimes arcane vocabulary—judgments asserted to be beyond scholarly debate.  Occasionally the scholarly impartiality of such experts was questioned due to their financial ties to dealers and auction houses.  At the same time, new critical approaches such as feminism, Marxism, post-colonial theory, and others—which questioned old hierarchies and the "authority" of the connoisseur (and even the "valorization of the object")—came into ascendancy in leading university art history departments.  As a result, art history split into two camps that have continued to diverge to varying degrees in methodology, terminology, and objectives, leaving "museum people" in one branch of the discipline and "academic art historians" in another.  Only recently, as many scholars have moved beyond theory, have there been signs that the two branches are beginning to intertwine again.

Art history seems poised to embrace a resurgence in connoisseurship, motivated in part by a "new formalism" in literary and visual studies, and also by a new interest in the way both Early Modern and Modern art markets were developed and the role therein of connoisseurs in establishing prices, developing concepts of authorship and authenticity, and promoting certain types of collecting and documentation.  We argue that future art historians will return to many of the traditional methods of connoisseurship (close study of objects in the original) and that these methods will benefit by utilizing the rigorous methods developed in computer vision and scientific image analysis.

Several of the techniques of traditional connoisseurship have counterparts in, and can be enhanced by, computer vision methods:
Close comparison of works  Computer methods can aid in revealing subtle differences between works (for connoisseurs to interpret) but more importantly quantify the differences objectively.  Such computer methods support diachronic studies of a given artist as well as studies of artistic influence.

  • Presentation of works in non-traditional ways Connoisseurs sometimes present art works in unusual ways, for instance inverting an art work, so as to allow the connoisseur to concentrate on style—brush strokes, shading, color, and so on—rather than on the subject matter.  Computer image methods, too, often represent images in unusual ways, including wavelets or fractal coefficients or multi-spectral images, thus allowing the connoisseur to concentrate on style rather than content. 
  • Brush stroke analysis Art scholars study brush strokes to identify artists as well as the number of hands in a single work; new computer methods use subtle statistical tests to similar ends.
  • Lighting analysis  Art scholars have employed cast shadow analysis, but new computer methods reveal lighting inconsistencies within a tableau, thus providing information about the studio conditions and artists' praxis.

Computer methods introduce new analytic tools to art scholarship as well:

  • New visual measures  Computer image methods introduce fractals and multifractals—visual features that may be difficult to discern by eye.
  • Computer graphics Computer graphics reconstructions of studios and tableaus allow scholars to explore "what if" scenarios to infer artists' working methods, including possible use of drawing aids.  Computer graphics can also reveal some properties of lighting in realist paintings better than can the expert eye.
  • Machine learning  Connoisseurs develop insight through extensive viewing of large numbers of works.  Computer methods permit the statistical learning of regularities extracted from large digital corpora of works, corpora that can be extremely large or specially selected.

Computer vision methods do not replace expert connoisseur judgments, of course, but rather enhance and extend them—much as a microscope empowers a biologist.  This talk will be directed primarily to both art scholars who will someday use these new methods and will address problems in different languages and methodologies in the different disciplines, all in order to foster interdisciplinary collaboration.  The talk presents a vision of new art historical methodology:  a rich and powerful merging of traditional methods—close readings and comparisons of works, textual analysis—with new computer vision analysis methods, all employed by experts fully versed in art historical questions and contexts.

 

Return to main page