The renown abstract painter Wassily Kandinsky has been a common muse for computer-scientists and programmers to illustrate the capabilities of technology. The purpose of this paper is to evaluate those efforts and demonstrate that digital imitation does not correlate with the increasing power of the computer.Since Kandinsky’s passing in 1944 numerous attempts to articulate the painter’s style have been made. This research focuses on the generative approach taken by Pocock-Williams at MIT, the machine learning approach initiated by the Tensorflow team at Google, and a mixed approach used by Zhang et al the University of Texas. These studies are published over a thirty year period from 1988 to 2018. During this time the computer itself underwent exponential enhancements. Each imitation synthesizes a different aspect of Kandinsky’s work. Pocock-Williams recreates the formal composition, the Tensorflow team interprets the texture of Kandinsky’s paintings, and Zhang et al emulate individual visual elements.Despite these efforts no output is clearly more in Kandinsky’s style than any other. This inability to perfectly capture the painter’s style with a computer further demonstrates Kandinsky’s genius. At a moment when technology is disrupting industries and creating new economies, the results from this study serve to question the cultural impact of computing for current and future creative practitioners.
Long abstract of your presentation
Main author information