完全按照所示保留上传的参考角色,包括面部身份、面部比例、眼型、鼻子、嘴唇、肤色、发型、发色、可见的发饰、服装、造型以及整体可识别气质…
分享链接 · https://openimages.ajiang.me/p/2083295015346782494/




composition · full-bodycomposition · multi-panelhuman_attributes · long-hairlighting · natural-daylightmood · playfulscene · studio
提示词(原文,逐字保留)
Preserve the uploaded reference character exactly as shown, including face identity, facial proportions, eye shape, nose, lips, skin tone, hairstyle, hair color, visible hair accessories, outfit, styling, and overall recognizable vibe. Do not invent or hardcode any appearance details that are not present in the reference image; infer all identity, clothing, hairstyle, and accessories directly from it. Create a high-quality vertical (4:5 or 9:16) mixed-style portrait featuring a realistic full-body (or nearly full-body) version of the character standing on one side of a clean wall, with a black hand-drawn doodle-shadow of the same character sketched directly onto the wall beside them. The doodle-shadow must be black sketch line-art only (not a real shadow, not a second realistic person, and not a full-color anime character) while clearly matching the character's hairstyle silhouette, accessories, outfit silhouette, and overall identity. The real person should remain photorealistic, stylish, and naturally posed, wearing the exact outfit and key styling details from the reference, with a cute, slightly confused, mildly embarrassed, playful expression as if wondering, "Why am I doing this with my shadow?" The doodle-shadow should be much more energetic, exaggerated, mischievous, and chaotic, surrounded by manga-style motion lines, hearts, stars, sparkles, and comic effects. For every generation, invent a completely new playful mischievous pose instead of repeating previous ones—avoid pointing poses, finger guns, repetitive standing poses, or simply having both figures point at each other. The real person should perform a believable, softer version of the pose while the doodle-shadow performs an over-the-top cartoon version of the same idea. Choose a fresh concept each time such as a playful idol pose, silly dance, leaning sideways with one arm overhead, giant heart pose, cheeky wink, hands near cheeks, dramatic ta-da, mock surprise, running in place, mischievous tiptoe, arms stretched apart, sneaking away, cute troublemaker, dramatic overreaction, goofy victory, playful balance, peekaboo, shy-but-mischievous, or overconfident pose, while always creating a unique variation rather than copying these examples. Keep enough space around both figures so the full pose is visible, with the doodle-shadow roughly the same height or slightly taller than the real person, resulting in a cute, playful, stylish, funny, mischievous, and social-media-friendly composition.
中文译文
完全按照所示保留上传的参考角色,包括面部身份、面部比例、眼型、鼻子、嘴唇、肤色、发型、发色、可见的发饰、服装、造型以及整体可识别气质。不要凭空捏造或硬编码参考图中不存在的任何外观细节;所有身份、服装、发型和配饰都应直接从参考图中推断。创建一张高质量的竖版(4:5 或 9:16)混合风格肖像,其中包含该角色的写实全身(或近乎全身)版本,站立在一面干净墙壁的一侧,墙上紧挨着角色手绘一个黑色的涂鸦影子。涂鸦影子必须仅为黑色线稿速写(不是真实阴影、不是第二个写实人物、也不是全彩动漫角色),但需清楚匹配角色的发型轮廓、配饰、服装轮廓和整体身份。真人部分应保持照片级写实、时尚且姿态自然,穿着与参考图完全一致的服装和关键造型细节,表情可爱、略带困惑、稍显尴尬、俏皮,仿佛在疑惑:"我为什么要和我的影子做这种事?"涂鸦影子则应更加充满活力、夸张、调皮和混乱,周围环绕着漫画风格的动感线、爱心、星星、闪光和漫画效果。每次生成时,都要全新发明一个完全不同的俏皮调皮姿势,不要重复之前的——避免指人的姿势、手指枪、重复的站立姿势,或仅仅让两个形象互相指着对方。真人应以更柔和、可信的方式演绎该姿势,而涂鸦影子则以夸张的卡通方式演绎同一概念。每次选择一个全新的创意,例如俏皮的偶像姿势、滑稽的舞蹈、单臂举过头顶向一侧倾斜、巨型爱心姿势、调皮的眨眼、手放在脸颊旁、戏剧性的"嗒哒"亮相、假装惊讶、原地跑步、调皮的踮脚、双臂张开、偷偷溜走、可爱的小麻烦制造者、戏剧性的过度反应、滑稽的胜利、俏皮的平衡、躲猫猫、害羞但调皮、或过度自信的姿势,同时始终创造独特的变体,而非照搬这些示例。在两个人物周围保留足够的空间,以便完整姿势可见,涂鸦影子与真人身高大致相同或略高于真人,最终呈现出可爱、俏皮、时尚、有趣、调皮且适合社交媒体的构图。
来源与署名
原文由 @DigitalManSteve 发布在 X:查看原推文。 本页逐字保留原文并提供机器翻译的中文解读;版权归原作者所有。
同工具的更多提示词
- 以上传的照片作为精确的模特参考。1:1 还原模特的外观:面部特征、眼型、鼻型、唇形、自然肤色、发长、发量、发质、发缝、发型、身体比例、自然面部表情、眼神以及整…@codewithhajra
- https://t.co/55yoRROR5F 夏日傍晚的街角咖啡馆,一盏暖色落地灯倚在藤编椅旁,夕阳穿过梧桐叶洒下斑驳光影。她侧身坐在高脚凳上,手肘支在桌面…@sada_ai
- 蓝色战车,蓝色战袍😈 Seedance 2.5 用这张图片能不能过审呢❓ GPT-image 2 prompt 👇 Photorealistic 3:4 fu…@johnAGI168
- 一位发光的赛博奇幻女神,身着华丽的黑色生物机械铠甲,其上刻有发光的青色、洋红和翡翠色电路。她有着飘逸的银蓝色长发,佩戴着花卉饰品和镶有宝石的耳环,双眸闪耀着粉…@LANDCASTER_92
- 年代穿越 / "同一地点的岁月流转" 固定一个人物或地点,仅改变时代: "1986 年的同一 [主体/地点]。准确更新服装、家具、汽车、标识和照片颗粒感以匹配…@aatif_j
- https://t.co/BbQzQjyrR0 使用参考照片中的面部,保留完全一致的身份、特征、表情、肤色和带有可见毛孔的自然肌肤质感。不要美化,不要出现AI…@sada_ai