Original document
GPT Image 2 · en
Scene:
A square cinematic office at night, seen from a calm three-quarter angle. The room has a desk, a soft desk lamp, tall windows with city lights, shelves, a wall mirror, and a polished floor reflection. Use deep navy shadows, warm amber lamp light, and controlled negative space so multiple figures remain easy to read.
Subject:
Show the same man at four life stages sharing the office: child sitting under the desk with a sketchbook, teenager leaning against the window, adult working at the desk, and elderly version reflected in the wall mirror. Preserve identity through the same broad brow, round eyes, slight gap-toothed smile, and a blue cardigan motif adapted across ages. Each version should feel physically present or reflected in the same room, not pasted from separate images.
Important details:
The emotional concept is a person surrounded by earlier and later selves while making a decision. Keep the office geometry coherent, with reflections and shadows obeying the room. Age differences should be clear but connected through facial structure, posture echoes, and the blue clothing motif. The composition should feel like one surreal photograph with cinematic polish.
Use case:
Mirror Worlds single study 5/8, testing same-person age consistency and mirror logic in an office scene without references.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no extra coworkers, no readable documents, no horror tone, no inconsistent facial traits, no broken mirror perspective, no cluttered desk pile.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.
