Галерея ИИ-промптов OpenZhenZhen · gpt-image-2.5 · image

Розово-зелёная дримкор-фотосессия моды

Pink and Green Dreamcore Fashion Editorial

«Розово-зелёная дримкор-фотосессия моды» — работа с использованием gpt-image-2.5 на OpenZhenZhen. Изучите исходный промпт и пример генерации для собственных проектов.

Розово-зелёная дримкор-фотосессия моды — AI image creation
Розово-зелёная дримкор-фотосессия моды

Промпт

Промпт на китайском

使用附带的 ref-10a-v1 图像作为唯一身份和姿态参考,将整张图转化为一幅统一的粉色与青绿色梦核时尚大片。保留同一位成年女性可辨认的身份、核心上半身姿势、主要轮廓、清晰可读的面部、大体画幅和关键主体关系。让她可见的服装与原姿势保持连贯。将场景重构为粉色沙漠向青绿色群山延展,笼罩在柔和的粉色薄雾中;这是从原文列出的梦境场景中选择的一种,不是九宫格或多场景要求。使用粉色天空与氛围雾气、玫瑰色皮肤高光、桃粉色空气、青绿色阴影结构、柔和湖绿色远景和薄荷色环境反光。整体对比柔和但辨识度鲜明。将 90 年代复古时尚摄影、Y2K 实验影像、独立音乐专辑封面氛围和超现实广告融合成一张经过明确艺术指导的画面。采用柔焦、胶片颗粒、低对比、柔雾光、轻微可控过曝、复古杂志印刷质感、微微偏移的色彩、干净留白和梦境般的层次。保持主体突出、面部清晰;构图统一,并具有精致、高级的视觉亮点。只创作一张图。不要复制参考图的九宫格,不要生成多个小图,不要添加无意义文字或水印,不要出现真实品牌标志,不要让五官崩坏或主体消失,不要让成品看起来像随机生成的 AI 图片,也不要只套用平面的粉色滤镜。原文没有要求精确的可见字样。

Промпт на английском

Use the attached ref-10a-v1 image as the sole identity and pose reference, and transform the whole image into one cohesive pink-and-turquoise dreamcore fashion editorial. Preserve the same adult woman's recognizable identity, core upper-body posture, main contours, clear readable face, broad framing and key subject relationships. Keep her visible clothing and pose coherent with the reference. Redesign the setting as a pink desert opening toward turquoise mountains beneath soft pink mist; this is one source-listed environment choice, not a request for a grid or multiple scenes. Use a pink sky and atmospheric haze, rosy skin highlights, peach-pink air, turquoise shadow structure, a muted lake-green distance and mint environmental reflections. Keep the contrast soft but distinctive. Blend 1990s retro fashion photography, Y2K experimental video, indie music-album-cover mood and surreal advertising into one intentionally art-directed frame. Use soft focus, film grain, low contrast, gentle mist light, slight controlled overexposure, retro magazine-print texture, subtly shifted colors, clean negative space and layered dreamlike depth. Keep the subject prominent and the face clear; preserve a unified composition with a polished high-end visual hook. Create one image only. Do not copy the reference's 3x3 grid, create multiple mini-images, add meaningless text or watermarks, show real brand logos, collapse facial features, make the subject disappear, make the result look like a random AI-generated image, or apply only a flat pink filter. No exact visible wording is required.

Открыть модель API

Больше промптов для изображений и видео · Создать изображение · Создать видео · Создать на ИИ-холсте