PROMPT RADAR / PROMPT人像时尚
个人资料 / 头像 - Sunlit Cafe Winking Portrait
A high-fidelity structured prompt for a photorealistic close-up of a young woman winking in a cafe, featuring delicate lace details and soft afternoon light.
@CyberTotal2026 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
Theme:
One eye winking by a green window
Subject:
A close-up chest-up portrait of a young woman wearing a {argument name="clothing" default="white lace-trimmed dress"} leaning her cheek on one hand and smiling with one eye closed at a wooden table in a {argument name="background" default="bright cafe surrounded by greenery"}. The figure's face and upper body are placed large from the center to the left, showing a lemon drink on the bottom right, a light pink device in the foreground, and trees outside the window.
Person/Expression:
A small, rounded oval face tilted slightly to the right, chin slightly tucked. The large dark brown eye on the left side of the screen looks forward, while the eye on the right is completely closed in a wink. Slender brown eyebrows, a small straight nose, and a friendly smile with glossy pale pink lips closed and corners turned up. Fair skin with pale pink cheeks. Long dark brown hair down to the chest, parted near the center, creating thin bangs and strands along the cheeks, with loose hair hanging long over both shoulders, especially in front of the chest on the right.
Clothing/Pose:
A warm white sleeveless long dress. Thin shoulder straps and a deep V-neck are trimmed with small matching lace, with fine gathers spreading from under the chest and delicate vertical wrinkles in the thin fabric. An ultra-fine gold chain with a small transparent pendant around the neck, and a {argument name="accessory" default="thin pale gold bracelet"} on the right wrist. Sitting leaning forward toward the table, supporting the right cheek with the palm and bent fingers of the right hand. The left arm is extended forward and cut off at the bottom left.
Background/Light:
The foreground is a bright wood-grain round table. At the bottom right is a transparent drink with sliced lemon and ice, a black straw, and a round rattan coaster. In the lower center is a triple-lens device in a pale pink case. The background consists of black window frames, large glass, dense green foliage, white rattan chairs, and wooden pillars and a terrace in the back right. Soft diffused afternoon light from outside the window at the top left evenly illuminates the face, hair, shoulders, and table, casting thin shadows under the cheeks and arms.
Composition/Camera:
3:4 aspect ratio. The camera is at a nearly horizontal eye-level slightly higher than the seated subject's eyes, a chest-up close-up capturing from the crown of the head to the bottom of the chest and the edge of the table. The figure is captured large, about three-quarters of the screen height, with the face in the center of the upper half, the hand supporting the cheek in the right center, and the chest at the bottom left. The drink is partially cut off at the right edge, and the device is partially cut off at the bottom edge. Sharp focus on the open eye and lips, with the background leaves softly blurred.
Texture/Style:
A high-definition photorealistic photo based on white, light wood colors, fresh green, and pale yellow. Naturally depicts fine skin, glossy lips, thin hair, lace and thin fabric, ice and lemon, wood grain, and the bokeh of leaves outside the window.
Negative:
Do not change the winking expression and the hand supporting the cheek; do not remove the lemon drink and the light pink device.
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 high-fidelity structured prompt for a photorealistic close-up of a young woman winking in a cafe, featuring delicate lace details and soft afternoon light.
- 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
- 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