台大心理系

Pictorial composition, understood as the arrangement of graphical elements on the plane, is typically associated with simple rules and heuristics, such as the golden number. Beyond these constraints, we believe that artworks are able to fix more fundamental regularities in their pictorial matter. We therefore develop the idea that every artwork of an artist represents a partial view of a higher-order object, hyper-compositional. Practically, we employ artificial neural networks, trained on more than 6000 personal abstract drawings, and we check the concordance of this artificial model with human perception through psychophysical experiments. The ambition of this research-creation project (PhD defended last November) is thus to propose scientific means to measure compositional aspects of art, to reveal intrinsic hidden dimensions, and to show the dynamic of graphical elements through new artworks.