
原文
GPT Image 2 · en
Scene:
Use the reference image as the identity source for Fitness Wellness Brand Portraits. Build a stronger mood-lighting portrait that tests the outfit under a second lighting logic Keep the camera logic commercially believable, with stable perspective, square-crop discipline, and light that supports facial recognition before style. The image should feel finished enough for client review, media rollout, or research comparison rather than like an unfinished test frame.
Subject:
Show the same fashion model as the reference, now wearing a dramatic but still commercial look under duskier, contrast-rich or spotlight-oriented light. Preserve the exact recognisable face, haircut base, body read, and age impression from the anchor image while allowing the requested lighting, wardrobe, pose, or environment change to do the storytelling. The variation should clearly connect to the topic brief Produce athletic and wellness-driven portrait campaigns with realistic body proportions, clean skin texture, and premium studio-commercial lighting..
Important details:
Preserve exact face / hair base / body proportions from the reference model. Preserve the face shape, body proportions, shoulder width, leg length impression, hair base, and skin-tone family while the wardrobe and mood shift. Keep the skin, hair, garment, and background rendering realistic and production-ready. This frame should feel like a true commissioned variation within one coherent portrait package, not like a new person or a disconnected mood board.
Use case:
Fitness Wellness Brand Portraits series study 8/9, lighting-shift fashion variation using the reference identity.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no nightclub cliché, no crushed detail, no fashion-week chaos, no duplicate people, no extra fingers, no broken anatomy, no face drift, no oversmoothed skin, no unrelated props.使い方
- パラメータを含む元のプロンプト全体をコピーします。
- 被写体、文字、構図を自分の要件に置き換えます。
- 対応する外部ツールで使用し、利用条件と結果を確認します。
開始前の確認
- 参考画像の要否を確認してください。互換モデルの表示は、そのモデルでの生成実績を意味しません。
- 名前、数値、画像内の文字を確認します。モデルの版で結果は変わります。





