Identity Preservation Subway Portrait
A specialized prompt for gpt-image-2 that locks facial identity from a reference image while placing the subject in a cinematic subway station setting.
Prompt
Use the uploaded image as the ONLY facial identity reference. Absolute face preservation mode. Strict face lock. Maintain 95–100% facial similarity and preserve the exact eyes, eyelids, eyebrows, nose shape, nostrils, lips, jawline, cheekbones, forehead, skin tone, ethnicity, age appearance, facial proportions, beauty marks, and natural skin texture from the uploaded face reference. Do not redesign, beautify, alter facial structure, blend identities, or generate a different person.
Recreate the exact scene shown in the reference image: a candid urban subway platform environment with a train on the left side, tiled station walls, cinematic shallow depth of field, natural station lighting, soft background blur, realistic motion atmosphere. Subject looking back over her shoulder toward the camera with a warm genuine smile. {argument name="hair color" default="Long dark hair"} visible from beneath a light pink {argument name="headwear" default="New York Yankees baseball cap"}, small gold hoop earrings, oversized black outfit, black shoulder bag. Natural candid street-photography aesthetic, authentic lifestyle moment, realistic skin texture, ultra-photorealistic DSLR quality, shallow depth of field, soft bokeh, cinematic color grading, highly detailed, 8K resolution.
Important: Keep everything exactly the same as the original image (pose, camera angle, subway platform, cap, clothing, lighting, composition, background, expression, and environment). Replace only the face with the uploaded face reference while maintaining perfect identity preservation and realistic integration
Open in Hitboard Original post