beardcoded

Generative painting with code forms the core of his practice, where he studies abstraction by developing algorithms that support and extend the process of painting.¹ In How much of one is in the other (2022), released on fx(hash), each minted edition generates a unique colour from its hashcode by holding one RGB channel constant while circling the other between 0 and 255, so that the image also reads as a graph of its own colour values. The title is conceptual as well as descriptive: the work reflects on how individual, differentiated selves are still bound together by shared human qualities.²

He has also used AI image models in his practice.³ Automimetic painting: Monalisa and the AI painters uses a variational autoencoder to render the Mona Lisa in the styles of dozens of painters, from Leonardo da Vinci to Wassily Kandinsky and Yayoi Kusama, as a way of studying how an AI model's approach to style differs from a human painter's.³ In Facing biases, he used a public face-generation model trained on datasets such as ImageNet to draw faces, finding the underlying model skewed toward some populations, such as Asian and Global South faces, being represented far less than others.⁴