PROMPT RADAR / PROMPT人像时尚

社交媒体帖子 - Golden Hour Fisheye Bedroom Portrait

A cinematic photorealistic prompt for creating an intimate fashion-editorial portrait of two adult women in a warm sunlit bedroom.

@VIBEQUIRKLABS 0 热度 0 浏览
社交媒体帖子 - Golden Hour Fisheye Bedroom PortraitORIGINAL VISUAL / 来源见文末
编辑先看

它为什么值得进入灵感库

它最有价值的部分,是把一个模糊想法推进到了可以被模型执行的程度。 先锁定人物身份与服装轮廓,再用光线、焦段和动作消除“普通写真感”。

完整视觉 Prompt

Create a photorealistic cinematic fashion portrait of two adult East Asian women standing close together in a cozy bedroom at golden hour, both turned away from the camera but looking back over their shoulders with calm, serious expressions. The woman on the left wears a fitted ribbed slip dress in {argument name="left dress color" default="pale powder blue"} with thin straps, long slightly tousled dark hair, soft natural makeup, and dewy skin; the woman on the right wears a fitted ribbed slip dress in {argument name="right dress color" default="deep red"} with thin straps and an open back, shoulder-length dark hair, natural makeup, and warm highlights on her face and back. Their poses are elegant and intimate, with one arm gently around the other's waist, emphasizing fashion editorial styling rather than explicitness. Set the scene in a small hotel-like bedroom with exactly 6 main visible background elements: an unmade white bed with pillows on the left, a dark headboard, a framed landscape photo above the bed, tall sheer curtains on the right, a bright window behind the curtains, and a small bedside table near the window. Use a wide-angle fisheye lens look with curved distortion and dark vignette edges, as if shot on a {argument name="camera lens" default="16mm fisheye lens"}. Lighting should be warm sunset light streaming through the curtains, creating soft shadows, glowing skin highlights, and a moody amber atmosphere. Compose in a horizontal 16:9 frame, close mid-shot from behind at waist-to-shoulder height, shallow depth of field, high dynamic range, realistic skin texture, cinematic color grading, no text, no watermark, no extra people.
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 cinematic photorealistic prompt for creating an intimate fashion-editorial portrait of two adult women in a warm sunlit bedroom.
  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:Nailai7981 · 收录于 2026年8月11日

访问原始来源

沿着这条线继续看