PROMPT RADAR / PROMPT人像时尚
社交媒体帖子 - Car Selfie with Green Drink and Sunglasses
A realistic portrait prompt of a young woman taking a selfie in a car while drinking a green beverage.
@Sairah_0 0 热度 0 浏览
ORIGINAL 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”。
01 / 原始方法
完整视觉 Prompt
A young woman is taking a selfie, holding her arm out toward the camera. She is sitting in the back seat of a car on black leather seats. The interior is dark, with a panoramic sunroof. Outside, it is sunny and trees are visible through the windows and sunroof.
She is wearing a {argument name="top style" default="white sleeveless ribbed turtleneck top"}, several gold chains with pendants, and stylish {argument name="accessory type" default="white thick-framed sunglasses"} with opaque black lenses.
Her hair is loose, voluminous, and straight, with a few strands falling naturally across her face. She is wearing brownish-pink lipstick.
In her hand, she is holding a transparent cup filled with a {argument name="drink color" default="green drink"} by the lid from above, with a white straw that she is drinking from. Rings are visible on her fingers.
Vertical 9:16 format.
02 / 复刻路径
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。
- 主体:A realistic portrait prompt of a young woman taking a selfie in a car while drinking a green beverage.
- 视觉方向:社区实测视觉提示词;按原提示词的光线描述执行
- 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
- 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