Model Leans on Taxi in Manhattan — gpt-image-2

Why this works

The full-body, eye-level framing keeps the model’s confident stance and complete outfit readable, while “leaning against a yellow taxi” gives the portrait a strong diagonal anchor. “Soft late-afternoon natural daylight with warm neutral tones” tempers the black, gray, blue, and white palette, and “multiple yellow taxis” adds controlled color repetition against the office-building background. “Shallow depth of field” isolates the subject while “realistic reflections on surfaces” preserves the crisp, photorealistic street texture.

FAQ

How do I make the model more visually dominant than the taxis and office buildings?

Replace “Manhattan-style urban street background with multiple yellow taxis and tall office buildings” with “one softly blurred yellow taxi and minimal background architecture,” and change “eye-level full-body” to “slightly low-angle full-body editorial portrait.” Keep “shallow depth of field” so the urban context remains secondary.

How do I shift this from a warm, confident editorial into a moodier night street scene?

Replace “soft late-afternoon natural daylight with warm neutral tones” with “cool blue-hour light with hard storefront highlights and deep shadows,” and replace “warm neutral city color palette” with “cool charcoal, cobalt, and sodium-orange palette.” Add “wet pavement with reflected neon” while retaining “realistic reflections on urban surfaces.”

How do I create a series of variations without losing the same fashion identity?

Keep the fixed subject details, including “pastel-lilac graphic crop T-shirt,” “black satin mini skirt,” “light denim bucket hat,” and “white-and-navy sneakers,” then swap only the setting phrase for “SoHo side street,” “rooftop overlooking Midtown,” or “subway entrance at street level.” Vary “confident candid pose” into “walking mid-stride,” “turning over one shoulder,” or “seated on the taxi hood” while preserving “photorealistic street fashion editorial photography.”

Learn the technique behind this

Related prompts