Restyle a photo as a dusk urban streetwear portrait

Read the story behind this prompt
You Asked for Golden Hour. The Face in the Edit Was Wrong.
Read the articleTap any variable to replace it with your own value:
AI prompt
You are a street portrait photographer editing a real photograph in an image model that accepts a reference photo. Task: restyle the attached photo as a moody dusk city portrait in modern streetwear, while keeping the person's face and hairstyle identical to the reference. Inputs: - [PHOTO]: attach a clear photo of the person - [CROP]: three-quarter, or head and shoulders - [SETTING]: e.g. a city side street, a crosswalk, a shopfront at dusk - [WARDROBE]: default a structured dark bomber jacket over a simple layered tee - [LENS]: default a 35mm cinema-lens look - [ASPECT RATIO]: e.g. 4:5 Steps: 1. Lock the face first. Keep the person's face, hairstyle and exact facial identity completely unchanged from [PHOTO]: eyes, nose, mouth, skin texture, hairline and expression. Do not beautify or reshape the face. 2. Move the scene to [SETTING] at dusk, with a moody cinematic street, soft neon and out-of-focus streetlights in the background. 3. Dress the subject in [WARDROBE]. Keep the head angle and body proportions consistent with [PHOTO] unless a new [CROP] needs a little more of the torso. 4. Match the light: natural evening skin tones, realistic shadows that fit the ambient street light, a gentle rim from a shop window or sign. No harsh flash unless the street lighting would produce it. 5. Shoot it with a [LENS] look, sharp focus on the eyes, shallow but believable depth of field, authentic photography rather than a fashion illustration. Constraints and checks: - The face and hair must match [PHOTO] feature for feature, including moles, freckles and natural asymmetry. - Negative cues: a different face, plastic skin, extra fingers, warped ears, heavy makeup, unreadable neon text, watermarks, daytime sun. - If an optional input is blank, choose a sensible default and state it. Never relax the face lock. Output: - One edited photograph at [ASPECT RATIO]. - A short checklist confirming the face and hairstyle still match [PHOTO].
Paste into: ChatGPT, Gemini, Photoshop Generative Fill