AI Art Styles
This entry documents how a generative image model rendered Abstract when the movement was named in a prompt. It forms part of a survey of 60 art movements generated in February 2024.
The images
Two subjects are held constant across the series: Lake Kenogamissi in Northern Ontario, and Times Square in New York City. Each is rendered twice. A thin prompt names the movement and nothing else. A thick prompt supplies a generated description of the movement’s visual characteristics.
The thick descriptor
The following description was generated by GPT-4 and supplied to the image model as the thick prompt.
Abstract art disregards realistic references, often conveying ideas via shape, form, color, and line. To create an abstract piece, emphasize the following: 1) Color: Use vibrant and contrasting hues to evoke emotion, 2) Shape and form: Deploy non-representative or non-objective shapes/forms to move away from realistic imagery, 3) Line: Employ various types of lines (straight, curved, thick, thin) to express movement or energy, and 4) Departure from reality: Strive for a non-literal representation, favoring creativity and individual interpretation.
About this movement
Background on Abstract is available at its Wikipedia entry. The images above are not offered as an account of Abstract as art historians understand it. They record what a commercial image model produced when asked for the style by name.
About this series
This entry is part of a survey, described in the series introduction. The full set of 60 movements is browsable in the Art Styles index. The survey used text-to-image generation, in which composition varies alongside the style itself.
