AI photo transformation takes a source photo and a text prompt describing the desired result, then generates a new image that preserves key identifying features from the source while applying the style described in the prompt.
The two inputs that matter
The model receives your uploaded photo and a detailed text prompt (in Footty's case, a description of a specific club's kit, crest, and stadium) — both inputs shape the final output together, not just one or the other.
Why the result still looks like you
Modern image-editing AI models are specifically trained to preserve facial identity while changing everything else — clothing, background, lighting — around it, which is why the output reads as 'you in a kit' rather than a randomly generated stranger.
Why the prompt does so much work
A detailed, specific prompt (exact kit colors, real stadium name, atmosphere description) produces far more consistent, realistic results than a vague one — this is why well-written prompts, not just a powerful model, are a real part of getting good output.
Does the AI 'know' what a real stadium looks like?
Modern image models are trained on enormous datasets that include countless real-world photos, including stadiums and sports photography, which is why naming a specific real stadium in a prompt produces recognizably accurate architectural and atmospheric details.