
Original document
GPT Image 2 · en
Scene:
A square low-light smartphone snapshot inside a small corner store at night, photographed from the snack aisle with imperfect aisle-shelf framing. Fluorescent ceiling light reflects on vinyl floor, coolers glow in the background, and the camera slightly underexposes the subject's face. The composition should feel like a quiet accidental photo taken by a shopper, with realistic phone noise and shallow motion blur.
Subject:
A completely fictional older design-icon celebrity-like person stands near the checkout holding a plain paper carton and a pack of gum with no readable markings. They wear a black turtleneck under a simple gray coat, wire-frame glasses, and worn canvas shoes, but their facial features are invented and not based on any real designer or public figure. The cashier is mostly obscured; a second shopper recognizes the subject with a subtle double take.
Important details:
Use ordinary retail clutter without readable product text: colored blocks, abstract labels, shelf geometry, plastic baskets, taped floor marks, and refrigerator reflections. The subject should feel famous through calm presence and visual contrast against the mundane store, not through any real likeness. Make hands, carton, shelf perspective, fluorescent color cast, and phone-camera artifacts convincing. Avoid sensational or invasive framing.
Use case:
Fictional Celebrity Snapshot single study 2/8, testing a believable invented design-icon corner-store snapshot.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no real celebrity likeness, no real person name, no claimed real event, no cigarettes, no brand packages, no news photo caption, no staged commercial lighting, no uncanny face.How to use
- Copy the complete original prompt, including any parameters.
- Replace the subject, text and composition with your own requirements.
- Use a compatible external tool; check its current access conditions and review the result.
Before you start
- Check whether a reference image is required. A compatible model label does not mean this example was generated with that model.
- Keep names, figures and visible text accurate; results can vary between model versions.





