
原文
GPT Image 2 · en
9:16 Vertical screen — Creative forced perspective portrait,{argument name="style" default="Korean idol style"} A single subject. Simulating the aesthetic of 35mm film, with soft, natural grain, slightly faded colors, slight overexposure, soft tones, low contrast, soft highlight overexposure, and an ethereal, dreamlike atmosphere. A minimalist interior scene by a window, a clean, light-colored background, and a simple everyday environment. Young.{argument name="subject" default="Korean female idols"} Extremely attractive, with a natural, minimalist makeup look, dewy and realistic skin texture, and subtle imperfections. Clothing:{argument name="outfit" default="A loose white shirt casually draped over the shoulders + shorts"} Soft fabric, loose cut, conservative yet subtly figure-hugging, no revealing elements. Hairstyle: Long, dark hair, slightly messy, with a natural, voluminous feel. Composition (key): Forced perspective illusion—the subject appears unusually small relative to nearby objects (e.g., in a cup, standing on the edge of a table, or interacting with a hand in the foreground), yet still realistic; a sense of proportion blurring is created through camera distance and alignment. Camera: Slightly low or eye-level angle, deliberate perspective distortion, foreground objects close to the lens, subject placed further away to create an illusion of size. Pose: Natural, relaxed posture, slight body curves, casual standing or slightly sitting. Expression: Soft, calm gaze, a slight smile or slightly parted lips, natural and unforced. Lighting: Diffuse natural window light, soft shadows, the light gently enveloping the subject. Atmosphere: Playful, slightly surreal yet grounded in reality, an intimate everyday moment with visual reversal. Quality: Hyperrealistic, with fine film grain, slight soft focus, imperfect composition, and natural flaws.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。





