Original document
GPT Image 2 · en
Edit the provided image while preserving the same face identity, shape, and facial features without altering age, ethnicity, or structure. Maintain a calm, relaxed expression with the subject not looking at the camera.
Subject: young woman (18–23) with soft feminine beauty, smooth glowing skin, natural texture.
Pose (strict): seated on a wooden chair, body angled sideways (45–70°), legs crossed naturally, upper body slightly leaning forward, head turned away from camera, gaze to the side. She holds a drink cup with a straw using both hands.
Framing: full-body vertical (9:16), head to shoes visible, centered but slightly offset for natural composition.
Outfit: light pink varsity jacket with white stripes, soft pink inner top, modest knee-length pleated skirt, white sneakers.
Accessories: beige newsboy cap, sunglasses (on face or cap), pink shoulder bag, small earrings.
Hair: neat low bun with soft loose strands.
Environment: outdoor flower shop street scene with pastel flowers (pink, soft tones), decorative plants, floral storefront.
Lighting: bright natural daylight, soft glossy skin highlights, balanced exposure, soft shadows.
Camera: low/frog angle (slightly below, looking up), 50mm lens, shallow depth of field.
Color grading: warm pastel palette (pink, cream), clean bright lifestyle aesthetic.
Quality: ultra-photorealistic, 8K detail, DSLR realism, natural skin texture.
Negative: front-facing, eye contact, close-up, cropped body, mini/short skirt, indoor scene, dark lighting, anime, cartoon, CGI, plastic skin, distorted anatomy.
--ar 9:16 --style raw --quality highHow to use
- Copy the complete original prompt, including any parameters.
- Replace the subject, text and composition with your own requirements.
- Use a compatible external tool; check its current access conditions and review the result.
Before you start
- Check whether a reference image is required. A compatible model label does not mean this example was generated with that model.
- Keep names, figures and visible text accurate; results can vary between model versions.