OpenZhenZhen AI 提示词画廊 · gpt-image-2.5 · image

街拍与插画双生

Street Photography and Illustration Twins

竖幅9:16上下对半对照：顶部保留真实街拍原片，底部将同一人物、穿搭与街景转绘为现代日系社论时尚插画；两部分保持裁切、比例、姿势和构图一致，不添加文字。

街拍与插画双生 — AI image creation
街拍与插画双生

提示词 / Prompt

中文提示词

使用上传的真实街拍照片作为来源，并将已配置的身份参考 ref-10a-v1 用于照片中的主要成年时尚人物。制作一张竖幅9:16对照图，上下水平分为完全相等的两半：顶部50%——保留原始照片、人物及其可辨识身份、发型、身体比例、姿势、动作、服装颜色与轮廓、鞋子、包袋、配饰和整体氛围。维持真实纹理、原始光线和环境感，只做细微的社论级色彩校正。底部50%——将同一人物重新绘制为精致的现代日系社论时尚插画，不要只套卡通滤镜或描摹照片。保持人物清晰可辨，并保留原始表情与个性；采用略经精修但自然拉长的比例、干净连续的手绘线条、哑光色块、微妙色调变化和柔和手作纸感。服装细节优先忠实呈现：层叠关系、肩线与腰线、衣裙长度、口袋、衣领、袖口、鞋履、包袋、眼镜、珠宝及其他可辨识配饰。面部可雅致简化，但避免过大的动漫眼睛、Q版比例、娃娃脸或夸张可爱表情。将周围建筑、店面、窗户、街道、树木、阴影与环境色块也转换成相同插画语言，不要移除场景。只简化次要背景细节；保留原构图、相机角度、光线方向和空间关系。使用克制的手绘细节，呈现时尚或生活方式杂志中可收藏的OOTD页面质感。下半部分避免写实渲染、光泽3D、严重水彩晕染、厚重油画纹理、过多线条、通用卡通风格或重新设计服装。顶部与底部在裁切、人物比例、姿势和构图上紧密匹配。两个部分各占画布恰好50%。不要添加文字、标题、标签、徽标、品牌名称、水印、边框、前后对照标签或装饰排版。

English prompt

Use the uploaded real street-style photograph as the source and configured identity reference ref-10a-v1 for its principal adult fashion subject. Create one vertical 9:16 comparison image with two exactly equal horizontal halves: TOP 50% — preserve the original photograph, person and recognizable identity, hairstyle, body proportions, pose, movement, outfit colors and silhouettes, shoes, bags, accessories and overall mood. Retain realistic texture, original lighting and atmosphere; apply only subtle editorial color correction. BOTTOM 50% — redraw the same person as a polished contemporary Japanese editorial fashion illustration, not a simple cartoon filter or traced photo. Keep the person clearly recognizable and preserve the original expression and personality; use slightly refined, naturally elongated proportions, clean continuous hand-drawn linework, matte color fills, subtle tonal variation and soft handcrafted paper texture. Prioritize faithful outfit details: layering, shoulder and waist lines, garment lengths, pockets, collars, cuffs, footwear, bags, glasses, jewelry and recognizable accessories. Simplify facial features tastefully without oversized anime eyes, chibi proportions, doll-like faces or exaggerated cuteness. Translate the same surrounding architecture, storefronts, windows, street, trees, shadows and environmental color blocks into the illustration language rather than removing the setting. Simplify only incidental background detail; maintain the source composition, camera angle, lighting direction and spatial relationships. Use restrained hand-painted details and a collectible fashion/lifestyle-magazine OOTD finish. Avoid photorealistic rendering in the lower half, glossy 3D, heavy watercolor bleeding, thick oil-paint texture, excessive linework, generic cartoon styling or redesigning the outfit. Match the top and bottom closely in crop, subject scale, pose and composition. Each section occupies exactly 50% of the canvas. Add no text, title, label, logo, brand name, watermark, border, before/after label or decorative typography.

查看 API 模型

探索更多图片和视频提示词 · 生成图片 · 生成视频 · 在 AI 画布中创作