AI Selfies is a series of self-portraits that were never photographed. Since 2024 I have used my own face as source material for generative image models. Trained on my photographs, the models have learned to imitate me, and the images they produce look like photographs while having no moment of exposure, no place and no day they were taken.
The project explores the relationship between photography and synthetic imagery. Every image is both likeness and invention: a loop in which the self is mirrored back as fiction. What remains of a self-portrait when the camera is replaced by a model? What does authenticity mean when an image of me can exist without me ever having been in front of a lens?
The portraits deliberately borrow the visual language of social media: the selfie, the casual snapshot, the curated self. In doing so they question the trust we still place in photographic images of people, at a time when such images can be produced endlessly and convincingly.
All images are made on my own computer, using local GPU inference with ComfyUI, Stable Diffusion, FLUX and custom LoRAs trained on my face. Each image is documented and signed with C2PA content credentials, so that its synthetic origin remains traceable. The fiction is the subject of the work; it is never meant to deceive.
Timo Schuster, Berlin, 2025–2026
