After Constable's Clouds: Toward A Machine Learning Paradigm for Studying Pictorial Realism

The further development of computer vision methods to compare and classify features in paintings, specifically those from the Barbizon, Realist, and Impressionist traditions.After Constable's Clouds will use computer vision to enhance art historical understanding of 19th-century Realism. The emergence of Realism in French landscape painting is often linked to the 1824 exhibition in Paris of John Constable's unidealized view of the English countryside, The Hay Wain. Viewers particularly noted the veracity of Constable's clouds. Indeed, our computational research shows that Constable's clouds are more closely modeled on the structure of actual clouds than those of his contemporaries, with French academician Pierre-Henri de Valenciennes a near rival. Valenciennes taught a generation of landscape artists, emphasizing the importance of plein-air sky studies, yet histories of French landscape tend to cast Constable as Realism's catalyst. After Constable's Clouds will test this historiography by using computer vision to classify and compare the clouds in paintings by Barbizon, Realist, and Impressionist painters with those of Constable and Valenciennes.