GPT Image 2.5 prompt for a two-person editorial portrait

Nailai7981 @VIBEQUIRKLABSOriginally posted 2026.10.14

Open original on X

GPT Image 2.5 : Image Type: Create a photorealistic editorial portrait. Composition: Knee-level view, front view. Subject: Two 20-year-old adult women. Woman 1: Has young beautiful Korean idol face, refined small face, clear bright eyes, polished youthful beauty, photogenic K-pop portrait balance, long straight hair with full Japanese bangs, smooth clean silhouette. Wears tight body-skimming upper-body fit, white cropped t-shirt, concise hem length, compact torso block, casual knit structure, ultra-low-rise waistband sitting very low on the hips, tight body-skimming lower-body fit, black low-rise dolphin micro shorts, contrast piping, contrast elastic waistband, front drawstring, side lace-up grommet detail, compact athletic hot-pants silhouette, styled with white ankle socks, clean casual cotton legwear, light green nike p-6000 running shoes, layered mesh panels, metallic-like overlay accents, early-2000s runner styling. Woman 2: Has young beautiful Japanese transparent face, soft natural features, clean gentle eyes, airy fresh beauty, subtle innocent portrait presence, long soft waves with defined wave shape, clean tapered ends, natural face-framing silhouette. Wears tight body-skimming upper-body fit, black tank top, sleeveless casual fit, clean sporty line, ultra-low-rise waistband sitting very low on the hips, tight body-skimming lower-body fit, dark grey micro athletic shorts, compact sporty cut, minimal hem length, lightweight activewear structure, styled with white knee-high socks, clean youthful legwear, neon pink running shoes, lightweight athletic shoe shape. Shared Expression: Both women look directly at the camera with gentle warm smiles, friendly shared presence, relaxed openness. Pose and Composition: Two women captured in an intimate best-friends selfie moment, casual affectionate body language, close social warmth, playful candid interaction, spontaneous hand placement and close candid crop. Their body posture is a natural asymmetrical body arrangement suited to the scene and shared action. Knee-level camera, level lens axis, legs emphasized when visible, 0-degree front duo view, both torsos face camera. Scene: The portrait takes place in world-scene architecture for the portrait: Tokyo, Shimokitazawa, Shimokitazawa vintage shop alleys stays visible around and behind the subject; the portrait subject remains the main subject while the selected district, architecture, street texture, signage, landmark fragments, and local spatial cues remain readable; portrait-friendly spatial anchor: narrow vintage-shop alley corner; local cues: Shimokitazawa narrow lanes, vintage clothing shops, small music venues, handmade signboards and posters; real-place guards adapted for the background: avoid Harajuku Takeshita Street exaggeration, keep low-rise local backstreet scale, avoid over-polished fashion mall look; keep clear spatial context, avoid a plain empty backdrop, do not let the world scene cancel the person, wardrobe, pose, or action. Lighting: Deep azure summer sky, saturated clean blue atmosphere, towering luminous white cumulus clouds, crisp cloud-edge detail, vivid blue-and-white daylight. Hard direct sunlight on the subject, crisp cast-shadow edges, strong side contrast, bright specular highlights. Camera Look: Inspired by Rinko Kawauchi, airy high-key image language. Airy high-key photography, pale softly overexposed daylight glow, cool pastel tones, delicate everyday realism, fragile quiet details. Fujifilm Provia film simulation, clean balanced color, transparent skin tones, bright natural contrast, crisp daylight-neutral rendering, smooth saturation response. Please generate four variations on the same theme. All four images should have different compositions, poses, and camera angles. Please generate them all at once as multiple options for comparison, rather than creating them one by one in sequence. Keep the character, art style, and clothing consistent; make significant variations in composition, pose, and camera angle. Return four separate images, not a single four-panel collage. --ar 9:16