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

写实角色三面板身份与服装参考表

Realistic Character Three-Panel Identity and Costume Reference Sheet

以固定绑定的人物参考图为唯一身份依据，制作横向三面板写实角色表：左侧头肩特写、中间全身正面、右侧全身背面；统一发型、服装、比例、影棚背景与光线，不添加文字或道具。

写实角色三面板身份与服装参考表 — AI image creation
写实角色三面板身份与服装参考表

提示词 / Prompt

中文提示词

以ref-10a-v1 固定人物身份参考图作为权威的人物身份与外观依据，创作一张横向16:9写实角色参考表。画面严格分为三个对齐的竖向面板，面板间用细直白线分隔，并在三处呈现同一位成年女性：左侧为较大的头肩特写，人物面向镜头、表情自然中性；中间为全身正面；右侧为全身背面。三个面板中的面部、成年外观、发型、发色与发丝质感、身体比例、服装、颜色、配饰和鞋履保持一致。准确保留参考中可见的细节；对不可见之处，只作保守且合理的同套装束与外观延续。中间和右侧面板以放松、对称的站姿展示从头到鞋的完整人物。使用无缝中灰色影棚背景、柔和均匀的摄影照明、自然接触阴影、真实皮肤纹理与毛孔、清晰发丝、细致面料纹理、锐利焦点和自然解剖结构。左侧特写大于任一全身面板；全身视图保持摄像机高度一致，留白均衡。最终严格生成一张包含且仅包含上述三个视图的成品图像。不得出现其他人物、额外面板、文字、标签、水印、道具、布景、装饰图形，也不得添加细白分隔线以外的边框。避免塑料感皮肤、过度磨皮、扭曲的手、多余手指、五官不一致、解剖变形、模糊和视觉伪影。采用所需的16:9构图，并使用图像工具实际支持的最高输出分辨率。

English prompt

Create one wide 16:9 photorealistic character reference sheet using ref-10a-v1 固定人物身份参考图 as the authoritative identity and appearance reference. Show the same adult woman in exactly three aligned vertical panels separated by thin straight white dividers: left, a large head-and-shoulders portrait facing camera with a natural neutral expression; center, a full-length front view; right, a full-length rear view. Keep the face, apparent adult age, hairstyle, hair color and texture, body proportions, outfit, colors, accessories, and footwear consistent across all panels. Preserve visible reference details; for unseen details, make only conservative, plausible continuations of the same outfit and appearance. The center and right panels show the complete figure from head to shoes in relaxed symmetrical standing poses. Use a seamless medium-gray studio background, soft even photographic lighting, natural contact shadows, realistic skin texture and pores, individual hair strands, detailed fabric weave, crisp focus, and natural anatomy. Keep the left portrait larger than either full-body panel, with balanced spacing and consistent camera height for the full-body views. Produce exactly one finished image with these three views only. No extra people, panels, text, labels, watermarks, props, scenery, decorative graphics, or borders beyond the thin white dividers. Avoid plastic-looking skin, excessive retouching, distorted hands, extra fingers, inconsistent features, warped anatomy, blur, and visual artifacts. Use the requested 16:9 composition and the highest output resolution actually supported by the image tool.

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