PROMPT RADAR / IMAGE插画三维

个人资料 / 头像 - Candid Japanese Street Portrait

Generates a realistic vertical smartphone-style portrait of a young Japanese woman seated on a quiet urban cafe street.

@daisukeonarimon 0 热度 0 浏览
个人资料 / 头像 - Candid Japanese Street PortraitORIGINAL VISUAL / 来源见文末
编辑先看

它为什么值得进入灵感库

这条内容值得留下的,不是表面的风格词,而是它提供了一条可以继续实验的画面路径。 轮廓语言、材质规则和世界观要互相支持,避免把互不相干的风格词堆在一起。

完整视觉 Prompt

Create an ultra-realistic vertical smartphone photo portrait of a young Japanese woman, {argument name="character age" default="early twenties"}, sitting outdoors on a quiet narrow city street beside a cafe or small storefront. She has a natural soft smile, clear skin, large dark brown eyes, and a short black bob haircut with light wispy bangs across her forehead. She is looking directly into the camera from a close three-quarter angle, shoulders slightly turned, relaxed and approachable. She wears a simple cream-colored knit cardigan with visible front buttons and soft oversized sleeves. The background shows an urban Japanese side street with pavement, building facades, a utility pole, overhead wires, a small road sign, and greenery from potted shrubs and slender trees beside a brick wall; include warm interior cafe light near the right edge. Use natural daylight, soft shadows, shallow but not extreme depth of field, realistic skin texture, candid iPhone-style framing, vertical 9:16 composition, face centered in the upper half, no makeup-heavy glamour styling, no text, no watermark. Overall mood: gentle, fresh, everyday street portrait with photorealistic detail.
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:Generates a realistic vertical smartphone-style portrait of a young Japanese woman seated on a quiet urban cafe street.
  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:daisuke onarimon · 收录于 2026年8月9日

访问原始来源

沿着这条线继续看