PROMPT RADAR / PROMPT平面设计

个人资料 / 头像 - Sunlit Seaside Winking Portrait

A photorealistic beachside fashion portrait prompt for creating a bright summer close-up with lens flare and pearl accessories.

@tinypia 0 热度 0 浏览
个人资料 / 头像 - Sunlit Seaside Winking PortraitORIGINAL VISUAL / 来源见文末
编辑先看

它为什么值得进入灵感库

先别急着照抄整段 Prompt:真正有用的是主体、空间和镜头之间的约束关系。 先规定信息层级和文字安全区,再谈视觉风格;无法落版的漂亮画面不是完整设计。

完整视觉 Prompt

Create a photorealistic sunlit outdoor portrait of a young {argument name="subject ethnicity" default="East Asian"} woman at a seaside resort, framed as a close-up from the chest up in a square 1:1 composition. She has long flowing {argument name="hair color" default="dark brown"} wavy hair tousled by a light breeze, glowing dewy skin, natural makeup, soft peach lips, and a playful expression with one eye winking and a gentle smile while looking directly at the camera. She wears a delicate beige floral lace camisole with thin straps and a small bow at the center, plus exactly 3 visible pearl jewelry pieces: one pearl stud earring, one short pearl necklace, and one longer layered pearl necklace with a small pendant. The background shows a bright blue sky with scattered white clouds, intense sun behind her upper-left hairline creating strong backlight, halo glow, and colorful lens flares, with a blurred ocean coastline, railings, greenery, and distant hills below. Use shallow depth of field, high-end fashion editorial photography, warm summer atmosphere, soft skin highlights, realistic hair flyaways, natural body posture leaning slightly toward the camera, and a dreamy vacation aesthetic. Avoid text, logos, extra people, distorted anatomy, over-smoothed skin, or cartoon styling.
查看原作者内容

不要一次改完所有变量

用 GPT Image 2 复刻时,先固定主体和构图,再逐次替换光线、色调或材质中的一个变量,这样更容易判断哪项描述真正起作用。

  1. 主体:A photorealistic beachside fashion portrait prompt for creating a bright summer close-up with lens flare and pearl accessories.
  2. 视觉方向:社区实测摄影提示词;按原提示词的光线描述执行
  3. 镜头与构图:按原提示词的镜头与景别执行;按原提示词的构图与画幅执行
  4. 交付约束:auto,遵循原提示词中的避免项;不要水印、乱码、低清、畸形手、错误透视或主体漂移。
负向约束

遵循原提示词中的避免项;不要水印、乱码、低清、畸形手、错误透视或主体漂移。

模型建议

GPT Image 2 / Grok Imagine / Midjourney / Flux。先用原始 Prompt 建立基准结果,再根据模型对自然语言、镜头运动或文字排版的能力做局部改写。

来源与编辑原则

保留作者,也保留判断。

本页收录公开案例并提供中文索引、结构化拆解和复刻建议,不冒充原作者作品,也不替代原帖上下文。

作者:Tiny · 收录于 2026年8月11日

访问原始来源

沿着这条线继续看