Automated Analysis of Paintings and Photographic Arts, James Wang
The talk will focus on two recent research efforts at Penn State on computerized analysis of paintings and photographic arts. Art historians have long observed the highly characteristic brushstroke styles of Vincent van Gogh and have relied on discerning these styles for authenticating and dating his works. In our work, we compared van Gogh with his contemporaries by statistically analyzing a massive set of automatically extracted brushstrokes. A novel extraction method is developed by exploiting an integration of edge detection and clustering-based segmentation. Evidence substantiates that van Gogh's brushstrokes are strongly rhythmic. We also found that the traits that distinguish van Gogh's paintings in different time periods of his development are all different from those distinguishing van Gogh from his peers. The talk will also describe a comprehensive system to enhance the aesthetic quality of the photographs. The system, named OSCAR, has been designed to provide on-site composition and aesthetics feedback through retrieved examples.
Online Collaboration in the e-Humanities, Hans Brandhorst
Eventually every library or museum that holds a copy of Dürer’s famous "Melencolia I" print will have put it online. As a result we shall have many more descriptions like the following:
“A despondent winged female figure holding a geometrical instrument surrounded by attributes associated with knowledge; symbolising melancholia...” or the next one:
“The winged personification of Melancholy, seated dejectedly with her head resting on her hand, holds a caliper and is surrounded by other tools associated with geometry, the one of the seven liberal arts that underlies artistic creation—and the one through which Dürer hoped to approach perfection in his own work...”
Whether any of them will ever mention all of its visible details – e.g. a hammer, a pair of scales, a bat holding a text scroll, pliers, a block plane, a file, a purse, nails, a set of keys, a bell, an hourglass – is doubtful. Whether descriptions like these will ever share a standardized vocabulary to express the interpretation of the print is even more doubtful. The only thing we know for sure is that there will always be room for one more interpretative description. New research will always find a new angle from which to look at it, and new ways of thinking lead to new interpretations of the same object’s visible features and invisible meanings. How a substantial corpus of images (> 500,000), indexed by subject in a common vocabulary, could be used as the starting point of a cumulative dialogue between researchers in the arts and humanities, will be the topic of my talk. It will also involve a demonstration of the interactive functionalities of the Arkyves website with the help of which researchers can start a dialogue, add new metadata to the original catalogue descriptions and comment on each other’s interpretations.
A New Presentation of Parmigianino's Self Portrait in a Convex Mirror, David Stork
In the past few years, rigorous computer image analysis has addressed a number of problems and even controversies in the history and interpretation of fine art, from authentication of putative Pollock drip paintings to claims that Renaissance painters secretly traced optically projected images nearly two centuries earlier than previously thought. This talk will present recent computer image analysis and computer graphics reconstruction of Self portrait in a convex mirror, an important early work by the Mannerist Parmigianino displayed in the Kunsthistoriches Museum in Vienna. The work was executed in Rome in 1524 on a circular wood support, convex, like the "barber's mirror" described in Vasari's Lives of the painters. The executed image and unique optics underlying this work means that the proper viewing position of this work is not perpendicular to the center of the work. To our knowledge there has never been even a single "proper" photograph of this painting in any art history book, monograph, or on the web. We present the first such presentation, and discuss the implications for both its museum display and for the history of art.
Computer Vision for Interactive Experiences with Art and Artistic Documents, Rita Cucchiara
Art is a marvellous product of human intelligence, designed for personal and collective experiences. For tens of centuries, the experience with art has been mostly passive: art has been contemplated by senses, and “processed” by the users’ personal priors. Now, every human experience with media is interactive-by-design, so experience with art: not only with digital art but also whichever artwork which has been digitalized or remains in its native form. Experiences with artistic documents, historical books, painting in museums are examples of artworks that need to be re- thought for interactive experiences.
Computer vision for art interaction will be discussed at two directions: from artworks to human and viceversa.
In one direction, the talk will present results in understanding the artistic visual content, to support experts, scholars, art-lovers in accessing artworks and web related data: experiences of multi-digitalization, i.e. transformation of digital artistic documents, with texts, pictures, and drawings into multimedia content by segmentation, content understanding and retrieval.
In the other, the talk will focus on understanding humans and their behaviours for augmented interaction: will discuss artworks recognition and gesture analysis from ego-vision wearable devices to answer users requests in interaction with art.
A few projects developed at Imagelab – UNIMORE with Italian museums and with Italian artistic documents will be presented.