NYC Rooftop Golden Hour — Thrift Flex Lookbook

Upload your own photo and get a stunning thrift-flex editorial shot on a golden-hour NYC rooftop — looks like it cost a full photoshoot.

Best tool ChatGPT (GPT-4o image)
Aspect ratio 4:5
Difficulty Intermediate

The prompt

Generate a high-fashion editorial photograph of [your selfie] styled in a vintage 1990s-inspired outfit — think oversized plaid flannel shirt tied at the waist over a white crop tee, high-waisted straight-leg light-wash jeans, and worn-in white Air Force 1s. The setting is a gravel rooftop in New York City during golden hour, with a hazy amber skyline and water towers visible in the background. The lighting is warm and backlit, creating a slight lens flare. The pose is candid and relaxed — mid-laugh, one hand adjusting a bucket hat. The overall mood is effortlessly cool thrift-store chic. Shot style: 35mm film photography with slight grain, Kodak Portra 400 color palette, warm highlights, lifted shadows.

Why it goes viral: Combining a personal face upload with a recognizable NYC aesthetic makes every viewer feel the content is attainable and aspirational at the same time.

How to use it

Upload a clear selfie image into the ChatGPT image generation interface.
Paste the full prompt, replacing [your selfie] with a reference to your uploaded image.
Request 3 pose variations in a single message to maximize output variety.
Download the best frame and add a minimal caption with outfit tag handles for engagement.

Pro tips

Upload a clean, well-lit front-facing selfie for the most accurate face preservation with GPT-4o image generation.
Specify the Kodak Portra 400 palette explicitly — GPT-4o responds well to named film stocks for consistent color grading.
Ask for 3 variations in one prompt by ending with 'Generate 3 versions with different poses: arms crossed, looking away, and mid-walk.'
If skin tone rendering feels off, add 'accurate warm-toned skin, true-to-photo complexion' to the prompt.

#goldenhour#90sthrift#NYCrooftop#KodakPortra#candideditorial#filmgrain#backlit