is a generative reinterpretation of the venus of willendorf figure . . .

a prehistoric figure that is closely tight to ideas of fertility, beauty, femininity, and the female body.

the work begins with an image of the original venus, combined with contributed documentation of body parts and written prompts. through generative image-making processes, these inputs are used to produce new bodily forms. rather than treating the figure as a fixed icon, seeding_venus approaches it as a site of production: a body repeatedly interpreted, classified, aestheticised, and reproduced.

within the cybersculpture, scrolling marquees display two kinds of prompts. in uppercase, they show labels associated with how the Venus of Willendorf appears within image datasets. in lowercase, they show prompts introduced to redirect or interrupt these inherited classifications. the work is an examination of how generative AI systems reproduce existing visual norms, especially those shaped by dataset labelling, cultural bias, and dominant standards of beauty and femininity on the internet.

when illuminated, seeding_venus initiates an ongoing procession of generated Venuses. each new figure emerges from the archival image, donated body, machine interpretation, and human intervention. it asks how bodies are seen, named, and made visible within contemporary image systems, and whether these systems can be unsettled from within.

the Venus of Willendorf has sometimes been speculated to be an early self-representation of a woman. seeding_venus extends this speculation into the present, asking what forms of self-image, embodiment, and authorship are possible when the body is mediated...


the work is inspired on the cd-rom: cyberflesh girl monster by linda dement (1995)
used gen AI systems: img2text (stable diffusion), LAION 400-m, midjourney, local stable diffusion
created for: model collapse, 2024

image 1 & copyright: tessa wiegerinck
image 2: ars electronica, tom mesic

generated bodies by seeding_venus