Exploring Cultural Innovations Interdisciplinarily with Artificial Intelligence (KIKI)
How were materials reflected, fictionalised, through their visual representation in the 14th and 15th centuries? What role did the representation of material properties and surface qualities play in visual communication strategies? In order to be able to examine development trends and patterns with regard to the depiction of materials in medieval painting and drawing over a broad source basis and to detect individual phenomena, the use of methods from the digital humanities is required.
In the project, on the one hand, we conduct analyses based on the manually generated scene graphs from the image database REALonline. The database is exceptional in providing annotations specific to the materials of the individual entities depicted in the image, since annotation data on depicted materials in medieval images is otherwise hardly available. On the other hand, AI-based texture classification methods are adapted in the project and applied together with semantic segmentation methods to be able to automatically recognise represented materials and determine the exact shape of the objects in the image. Based on the results, we will evaluate the application of visual computing methods for a distant viewing of material representations across large image corpora – being able to take into account factors such as the position and proportion of the materials in the images or comparisons between similar textures. The art historical case studies will focus on works from the 14th and 15th centuries from the area of the Prince-Archbishopric of Salzburg and neighbouring regions.
Funded by
Master of Laufen, St Helena finding the true cross (detail), ca. 1440/45, Graz, Universalmuseum Joanneum, REALonline 000671