Core technique
The prompt uses a reference-image separation strategy: identity is locked to the uploaded personal photograph while visual characteristics are transferred selectively from the reference. Composition is a close-up to medium close-up three-quarter portrait, with the eyes positioned around the upper-middle frame and the face kept as the focal point. A simulated 85mm full-frame portrait setup, wide soft light, restrained neutral colors, and shallow depth of field reinforce the editorial photography treatment.
Recommended base models
GPT-Image-1 — well suited to identity-preserving edits and detailed natural-language constraints. Flux — strong choice for photorealistic fashion styling, realistic materials, hair, and lighting. Mid journey — useful for the editorial aesthetic and composition, but stricter identity preservation may require more careful reference handling. SDXL — capable of detailed photographic rendering, though maintaining the exact face while changing extensive styling can require additional identity-control techniques.
Suggested negative prompt
different person, altered identity, face replacement, generic AI face, changed face shape, distorted eyes, crossed eyes, unnatural eyebrows, deformed nose, oversized lips, plastic skin, waxy skin, excessive skin smoothing, airbrushed face, porcelain skin, fake beauty filter, CGI appearance, cartoon, illustration, 3D render, uncanny face, excessive makeup, exaggerated contour, unnatural smile, distorted teeth, malformed ears, deformed neck, extra fingers, missing fingers, distorted hands, duplicated jewelry, floating jewelry, unrealistic fabric, messy clothing, harsh shadows, oversaturated colors, artificial glow, excessive sharpening, low resolution, blur, noise, artifacts, text, watermark, logo, cropped head, cropped hair, extreme close-up, wide-angle distortion, unnatural proportions