GPT Image 2AI 提示词
Anime Girl on Stolen Crescent Moon — Sora 2 Prompt
Generate a dreamy vertical anime illustration of a girl sleeping on a giant crescent moon above a gothic town, perfect for storybook or album art.

提示词 · EN
Create a dreamy vertical anime fantasy illustration of argcharacter description: a petite teenage girl with dark brown hair in a tousled side ponytail sleeping curled up on an enormous cratered crescent moon floating above a European gothic town at night. She wears {argument name="outfit style" default="a black and navy Victorian military-lolita outfit with gold trim, short layered skirt, capelet, gloves, lace details, ribbon bows, and black ankle boots"}, with a small ornate hat and crescent-shaped gold accessory in her hair. Her eyes are closed, expression peaceful, hands tucked near her face as if hugging the moon, legs folded comfortably along the lower curve. The crescent moon is large, pale ivory, rough and cratered, with 6 visible dangling gold chains and charms hanging from its lower edge: two star charms, one ornate diamond-shaped celestial pendant, and three small decorative drops. In the upper left sky, show a second bright full moon with a crescent-shaped bite missing from its right side, implying the sleeping girl has stolen a piece of the moon. The background is a deep cobalt-blue starry sky over a fairytale city of steep rooftops, lit windows, clock tower, spires, and distant castle silhouettes. Frame the scene as if viewed from an open window or balcony, with dark window edges on both sides, hanging green vines and leaves at the top, soft purple flowers and roses near the bottom, and a warm lantern glowing in the lower right foreground. Use cinematic moonlight, soft rim lighting, delicate painterly anime rendering, high detail, romantic magical atmosphere, depth of field, rich blues and golds, vertical composition, no text.
关于这个提示词
Generate a dreamy vertical anime illustration of a girl sleeping on a giant crescent moon above a gothic town, perfect for storybook or album art. 它更适合作为 GPT Image 2 的概念艺术起点:先保留画面结构、主体关系和风格约束,再替换成你的品牌、人物或场景。
使用时优先替换 character description、outfit style 等占位符,并保持镜头、构图和材质描述的顺序不变。这样更容易获得稳定的画面结果,也方便继续做多版本对比。
如何使用这个提示词
- Copy the full prompt text from the page.
- Open your preferred AI image generator (e.g., Sora 2, Midjourney, DALL·E).
- Replace the
{argument}placeholders with your own character and outfit descriptions. - Paste the customized prompt into the model's input field.
- Generate the image and adjust settings like aspect ratio (vertical recommended) if needed.



