In this Theory Lab we will look closer at the concepts of ‘abstraction’ and ‘style’, utilised to describe specific painterly qualities within art history. Naja Grundtmann will initiate the discussion by presenting the role of ‘style’ in convolutional neural networks, and by demonstrating how the concept of abstraction reveals the inability to separate content and style, thus countering the idea of Style Transfer in convolutional neural networks. This may lead to discussions on how systems such as Stable Diffusion and DALL·E 2 mobilise the art historical notion of style in a manner that radically challenges our most established and conventional definitions of the term. Textual point of departure is Chapter 2 from Naja Grundtmann’s PhD-dissertation (distributed upon signing up).

Please sign up by accepting the Outlook invitation or by mail to lottephilipsen@cc.au.dk.