Typologies 01: Faces

Developed using the open-source Stable Diffusion XL text-to-image model, these synthetic images have been fine-tuned with five custom-trained Low-Rank Adaptation (LoRA) models, each of them programmed with images from different decades drawn from the Polygoon Journaal.

Using only a few basic prompts, we've produced a typology of 6475 images, organised into an x/y grid representing emotions ('x') and roles ('y') without providing any further context. This encourages the custom-made AI model to make creative choices and envision a parallel, artificial Polygoon Journaal.