Ghosts

These works have been made using stable diffusion models that have been fine-tuned on data that doesn’t have people in. Usually this would wipe out the ability for a text-to-image model to generate images of people. But when prompting these models to generate images of super-models they maintain their likeness and the photographic qualities of high-end fashion shoots, but have lost all understanding of anatomical relationships. Super-models, being some of the most photographed people on earth, remain as ghosts encoded in the weights of the model when everything else has been forgotten.