
Díptico geométrico de fotos en estilo cubista
【构图与原图】生成一张3:4竖版上下双拼图,上下区域高度约1:1。输入为单个人物或单只宠物的参考照片。上半部分直接呈现输入原图,保持照片里的主体、姿态、神态、服装配饰、背景与光线;仅等比缩放安放,比例不合时补白,不拉伸、不重新绘制、不裁掉关键内容。下半部分只呈现同一主体的风格转译,纯白背景#FFFFFF,主体居中、视觉中心稳定、留有呼吸空间,不复制照片背景。照片是半身或近景时,下方对应同样的取景范围,完整容纳该范围内的主体,不硬补全身。照片是全身时,下方完整保留全身。 【下半部分|立体主义转译】创作「立体主义 Cubism」视觉语言的插画,提取该艺术运动的形体、色彩与构成原则,转译为现代可读的插画。使用多视点、分析性平面切分、相互嵌合的几何面和重组轮廓,将同一主体的正面与侧面信息组织在一个连贯形象中;几何结构承担空间分析,不以低多边形3D、玻璃碎片滤镜或随机三角形替代。人物的鼻梁、眼睛、脸颊、发型、服装与关键配饰可发生平面错位,但识别点和衣着结构仍清楚,不能额外长出五官或肢体。动物依据原耳形、额头、鼻口、躯干与毛色分区进行几何重构,保留物种和特定花纹的空间对应,不把宠物改造成机械动物。以原照片主色为锚点,结合赭石、灰蓝、炭黑或克制的辅助色形成清晰关系;有限的纸面、笔触与拼贴感只用于主体内部,保持单一完整主体和纯白背景,不出现印刷文字。 【中间衔接】下半区顶端用少量相互嵌合的几何面形成节奏,不切碎、不覆盖上方原图,不用粗黑横线。 【一致性与限制】人物保留脸型、五官关系、发型轮廓、肤色或其明确的风格化对应、服装结构、配饰、姿态与神态;动物保留物种、耳形、眼睛、鼻口、毛色分区、花纹走向、体型与项圈。允许所选风格需要的比例、材质和形体转译,但上下必须能认出同一主体。画面自然、有手作感,线条、色块、明暗、材质与纹理服务于同一风格,避免过度光滑、无意义细节堆砌、随机装饰、错误肢体、额外主体和混杂画风。除输入原图本来存在的文字外,不新增任何文字、字母、数字、标语、Logo、品牌名称、水印、边框或UI;下半部分去除参考图中的文字和Logo。每次只生成一张含上下两区的完整双拼图,不生成多主体示例拼图,不把人物版和宠物版拼在一起再裁切。
[Composition and Original Image] Generate a 3:4 vertical split-screen image, with top and bottom sections roughly 1:1 in height. Input is a reference photo of a single person or single pet. The upper half directly displays the input original photo, maintaining the subject, pose, expression, clothing, accessories, background, and lighting; scaled proportionally, padding with white if aspect ratio differs, no stretching, redrawing, or cropping critical elements. The lower half only shows a stylistic translation of the same subject on a pure white background #FFFFFF, centered with stable visual focus and breathing room, without copying the photo background. If the photo is half-body or close-up, match the same framing below without forcing a full body. If full body, retain full body below. [Lower Half | Cubism Translation] Create an illustration using the visual language of Cubism, extracting the art movement's form, color, and compositional principles translated into a modern readable illustration. Use multiple viewpoints, analytical planar division, interlocking geometric facets, and reorganized contours, organizing frontal and profile information into a coherent image; geometry serves spatial analysis, not low-poly 3D, shattered glass filters, or random triangles. Facial features, hair, clothing, and accessories may shift planes, but key identity points remain clear without extra limbs or features. Animals are geometrically restructured by ears, muzzle, torso, and fur patterns without turning into mechanical creatures. Anchor colors to the photo, combined with ochre, slate blue, charcoal, or restrained accents. Subtle paper, brushwork, and collage textures stay within the subject on a pure white background without printed text. [Middle Transition] A few interlocking geometric facets at the top of the lower zone create rhythm, without fragmenting or overlapping the top photo, and without thick black lines. [Consistency and Restrictions] Maintain identity, facial structure, expression, and clothing/fur patterns between top and bottom. Natural, handcrafted feel with unified style. No extra text, logos, or UI added.
| Modelo | GPT Image 2 |
|---|---|
| Proporción | 3:4 |
| Estilo | 创意玩法 |
| Autor | DAAI (@daaihq) |
| Post original | Ver en X |
Díptico vertical que transforma fotos de personas o mascotas en ilustraciones cubistas con planos analíticos y geométricos, ideal para diseño de arte e IP.
Cómo usar este prompt
- Copia el prompt original.
- Pégalo en GPT Image 2.
- Genera con la proporción 3:4.
Posted on X by @daaihq. PromptGallery only indexes it. Copyright stays with the author.








