我使用 GPT Image 2 创建了这张图片。

I created this image using GPT Image 2. This was the prompt I used: "Create a high-end Korean fashion lookbook featuring the same Japanese woman across all six looks, with strong identity consistency throughout the entire image. Her facial features, age, ethnicity, hairstyle, body proportions, and overall presence should remain unchanged in every panel, making it unmistakably the same person wearing different outfits. Arrange the composition as a clean and elegant 2×3 grid with six equal-sized full-body panels. Maintain consistent model scale, clear spacing between panels, and balanced editorial symmetry. At the top center, include refined luxury serif typography reading: 'LOOKBOOK — The Edit • Vol. 01'. Each panel should showcase a different outfit and natural fashion pose: 1. Beige oversized blazer, white top, wide-leg trousers, and heels. 2. Cropped leather jacket, fitted top, cargo pants, and luxury sneakers. 3. Black fitted evening dress with sophisticated accessories. 4. White oversized shirt paired with denim shorts. 5. Cream knit sweater with a pleated midi skirt. 6. Monochrome tailored jacket and matching skirt set. Give each look a unique standing pose and subtle change in expression while maintaining the same model identity and luxury campaign energy. The body language should feel confident, polished, and suitable for a premium department-store fashion catalog. Use a seamless warm ivory-beige cyclorama studio background with no props or furniture. Lighting should be soft, diffused, and inspired by Vogue Korea editorials, creating realistic shadows, natural skin texture, visible pores, and beautifully rendered fabric details. The final image should feel like ultra-premium commercial fashion photography with HDR quality, photorealistic detail, and 8K-level sharpness. Prioritize strict identity preservation, full-body visibility in every panel, clear panel separation, consistent proportions, and a sophisticated luxury presentation. Avoid: identity changes, different faces, face swapping, cropped figures, overlapping subjects, extra limbs, anatomical errors, cluttered backgrounds, props, furniture, logos, watermarks, anime styling, CGI, illustrations, overly smooth plastic skin, blur, or low-resolution output."

我使用 GPT Image 2 创建了这张图片。 这是我使用的提示词： "创作一本高端韩国时尚造型集（lookbook），在全部六个造型中呈现同一位日本女性，整张图像保持极强的身份一致性。她的面部特征、年龄、族裔、发型、身体比例以及整体气质必须在每个面板中保持完全不变，使人在不同服装下也能毫无疑义地认出这是同一个人。 构图采用简洁优雅的 2×3 网格，由六个等大全身体面板组成。保持统一的模特比例、清晰的面板间距，以及均衡的编辑对称性。在顶部中央加入精致的奢华衬线字体标题：'LOOKBOOK — The Edit • Vol. 01'。 每个面板应展示不同的服装与自然的时尚姿态： 1. 米色 oversized（超大廓形）西装外套、白色上衣、阔腿长裤与高跟鞋。 2. 短款皮夹克、修身上衣、工装裤与奢侈运动鞋。 3. 黑色修身晚礼服，搭配精致配饰。 4. 白色超大衬衫搭配牛仔短裤。 5. 米白色针织毛衣搭配百褶中长裙。 6. 单色（monochrome）定制西装外套与同款半裙套装。 在保持同一模特身份与奢华广告大片能量的同时，为每个造型赋予独特的站立姿态与微妙的表情变化。肢体语言应自信、利落，适合高端百货公司时尚目录的呈现方式。 使用无缝的暖象牙米色影棚背景纸（cyclorama），不要任何道具或家具。光线应柔和、漫射，灵感来自《Vogue Korea》时尚大片，营造逼真的阴影、自然的皮肤纹理、可见的毛孔，以及精美呈现的面料细节。最终图像应呈现超高端商业时尚摄影的质感，具备 HDR 画质、照片级真实细节与 8K 级清晰度。 优先确保严格的身份一致性、每个面板中的全身可见性、清晰的面板分隔、统一比例，以及精致的奢华呈现。 避免：身份变化、不同的面部特征、换脸、身体被裁切、主体重叠、多余肢体、解剖结构错误、杂乱的背景、道具、家具、标志（logos）、水印、动漫风格、CGI、插画、过度平滑的塑料皮肤感、模糊或低分辨率输出。"