PROMPT RECORD图像记录

「咖啡馆里一位美丽的女人，电影感」 效果：面部毫无特色，光线平淡，没有场景感。AI替你填上了所有你没做的决定——相机、镜头、光线、景深、服装、色彩调校。它全都选错了。 同一个场景。11层提示词...

中文说明

「咖啡馆里一位美丽的女人，电影感」 效果：面部毫无特色，光线平淡，没有场景感。AI替你填上了所有你没做的决定——相机、镜头、光线、景深、服装、色彩调校。它全都选错了。 同一个场景。11层提示词： 相机：Leica SL3，搭配 Leica APO-Summicron-SL 50mm f/2 镜头，光圈 f/2.0。光学滤镜：Tiffen Black Pro-Mist 1/4，营造微妙的光晕并柔化高光。 胶片：Kodak VISION3 250D，呈现丰富自然的日光肤色质感，细颗粒。 主体：一位 28 岁的越南女性，直黑发齐肩，穿着低饱和度的鼠尾草绿亚麻衬衫，佩戴小巧金色圆圈耳环。除淡唇彩外无明显妆容。 环境：一家小巧的巴黎风情街角咖啡馆，深色木桌、锌制吧台、一扇临街大窗。一面墙上贴着褪色的碎花壁纸。桌上放着一杯...

原始 Prompt

"A beautiful woman in a cafe, cinematic"

Result: Generic face, flat lighting, no sense of place. The AI filled in every decision you didn't make. It chose the camera, the lens, the lighting, the depth of field, the wardrobe, the color grade. It chose wrong.

Same scene. 11-layer prompt:

Camera: Leica SL3 with Leica APO-Summicron-SL 50mm f/2 at f/2.0. Optical filter: Tiffen Black Pro-Mist 1/4 for subtle bloom and softened highlights.

Film stock: Kodak VISION3 250D for rich, natural daylight skin rendering with fine grain.

Subject: A 28-year-old Vietnamese woman with straight black hair cut to her collarbone, wearing a muted sage linen shirt and small gold hoop earrings. No makeup visible except a light lip tint.

Environment: A small Parisian-style corner cafe with dark wood tables, a zinc bar counter, and a large street-facing window. Faded floral wallpaper on one wall. A half-finished espresso on the table.

Lighting: Natural daylight from the large window camera-right as the key light, creating a soft directional wash across her face with a gentle shadow on the far cheek. No artificial fill. The ambient warmth of the cafe interior provides deep warm tones in the shadows.

Composition: Medium shot, subject on the right third of frame, looking down at the cup with a slight smile. The window provides leading lines from the upper right.

Mood: Quiet, analog, lived-in. Warm midtones with slightly desaturated greens and browns. The Black Pro-Mist gives a nostalgic softness to the highlights.

That prompt produces an image that looks like a page from Kinfolk magazine. Not a stock photo. Not a template. A photo that feels like it was taken by someone who was there.

That difference — generic versus specific — comes down to how many layers you control.

I mapped the entire system into a formula. 11 layers. Each one with examples, before/after comparisons, and the exact syntax that triggers the AI to execute rather than guess.

Worth $97. Free right now if you:
- Like this post
- Comment "Prompt Hack"
- Follow @ArefAichemist

#AIArt #AIPhotography #CinematicAI #PromptHacking #CreativeAI #AITools