An autoethnographic practice-based study : exploring the use of Generative Adversarial Networks within the working processes of printmaking

Abstract

This research explores the creative effects of including Generative Adversarial Networks (GANs) in the printmaking working process. A GAN is an artificially intelligent computer model trained to mimic the abstract properties of a given dataset. The GAN generated images are used to create new prints, which are then used as data on which the GANs can re-train. The resultant body of prints yields a unique perspective that can be used as part of a feedback loop within the creative generative process. Printmaking in Western and South African art has been contested and regarded as a ‘non-art’. Mixed-media and hybrid printmaking are mitigation strategies which address this. Incorporating GANs falls within the remit of hybrid printmaking. Until now their influence on the creativity that drives the printmaking process remained largely unexplored. This research provides a basic understanding of GANs, with a focus on their limitations. GANism (art made using the medium of GANs) is studied in order to further understand the application of GANs within the art making process. A review of works by contemporary GANist artists Mario Klingemann (b.1970); Jake Elwes (b.1993); the French collective, Obvious; Tom White (b. unknown) and Anna Ridler (b.1985) has been conducted. An autoethnographic practice-based research methodology focused on the generative and explorative phases of the printmaking processes is conducted. The study includes reflective methods and documents a process-driven approach to using GANs as an artistic tool within the working process of printmaking. This research contributes to the existing hybrid printmaking and GANism scholarship by presenting an application of GANs as a digital tool within printmaking.