Sam Leach Machine-assisted river

Machine-assisted river has the hallmarks of a 17th-century Dutch landscape painting – a winding river and a twilight sky contrast with leafy foreground shadows. But this idyll is disquieting; foliage disintegrates into puffs of cloud and the river flows upwards.

Sam Leach’s starting point for this work was a machine-learning algorithm known as a generative adversarial network. Leach feeds the algorithm diverse scenic imagery, from Australian bushland to historical paintings, which generates new composite landscapes, based on what was input.

Leach works with the digital flaws and inconsistencies of the outcome to play with how we imaginatively construct and romanticise nature. He explains, ‘I am interested in how historical representations of landscape influence our understanding of the world around us and how we often reshape the world to resemble those idealised images.’

Leach won both the Wynne and Archibald Prizes in 2010; this is his third time in the Wynne. He is also the subject of a portrait by Keith Burt in this year’s Archibald.