OpenZhenZhen AI プロンプトギャラリー · gpt-image-2.5 · image

インタラクティブな古代美人キャラメイクのプロンプト

Interactive Ancient Beauty Character Creator Prompts

「インタラクティブな古代美人キャラメイクのプロンプト」は gpt-image-2.5 を使用した作品です。OpenZhenZhen で元のプロンプトと生成例を確認し、創作の参考にできます。

インタラクティブな古代美人キャラメイクのプロンプト — AI image creation
インタラクティブな古代美人キャラメイクのプロンプト

プロンプト

中国語プロンプト

以用户上传的参考图作为主要可见成年女性的面部身份依据。运行一个交互式古代美人捏人游戏。少解释，每次选择后立即保存状态，一次只展示一个项目。 流程严格按顺序进行：①朝代 → ②身材 → ③气质 → ④发式 → ⑤服饰 → ⑥妆容 → ⑦场景 → ⑧神态 → 完成。开始时只展示①朝代。每次选择后立即保存并推进，不询问是否继续，也不重复确认。 ①至③各提供 6-10 个紧凑视觉卡选项，包含编号、名称和不超过 10 字的短解析，以 2-3 列显示。朝代：先秦、秦汉、魏晋南北朝、隋唐、五代十国、宋、辽金元、明、清。身材：娇小纤秀、纤细柔婉、自然匀称、高挑修长、柔婉丰润、丰姿婀娜。气质：清雅、温婉、明艳、灵秀、端庄、妩媚、清冷、娇憨。 ④根据朝代提供 6-8 种有历史依据的女性传统发式。⑤根据朝代提供 6-8 种有文物、绘画、考古或文献依据的女性传统服饰形制；禁止现代新中式、泛中国风和影视架空服饰。⑥提供 6-8 种有历史依据的女性妆容。⑦提供 6-8 种符合该朝代建筑、生活方式和社会环境的女性生活场景。⑧根据当前全部设定动态生成 8 种明显不同的神态，不使用固定神态选项池。 联动规则：④至⑦必须与①朝代一致；辽、金、元分开处理，不得混合。修改①时只重新处理④⑤⑥⑦，②③保持不变。修改④至⑦任一项后检查后续设定，若冲突则重做受影响项目，并始终根据最新①至⑦重新生成⑧。 操作规则：随机只改变当前未锁定项目；换一个重新提供当前项目选项；锁定后限制随机、换一个和全部随机；明确使用“修改设定”时可修改已锁定项目；返回回到上一项目并保留不受影响的状态；全部随机只随机未锁定项目并从①开始；重新捏人清空全部设定。支持“换成明代”“发式换一个”“改成浅笑”等直接语义命令。无法点击时立即降级为编号或名称选择。 除 8 个捏人项目外，AI 可自动补全动作、手势、发饰、首饰、颜色、纹样、材质、鞋履、道具、人物关系、构图、镜头、光影、时令和天气，但必须符合选定朝代与人物设定。时令：春、夏、秋、冬；天气：晴、阴、薄雾、细雨、雪。 固定人物：成年女性、自然身形、比例协调、真实肤质。固定摄影：手机生活抓拍 × CCD 直闪、原生数码照片、9:16 竖幅；摄影风格不是选择项目，也不改变历史语境。 完成⑧神态后，必须依次输出：完整角色设定；当前单一场景；最终图片提示词，完整整合 8 项设定与所有自动补全细节；最后加入签名 ● DeepBlue，其中 ● 必须是纯 #0B3D91 深蓝色实心圆点，DeepBlue 使用白色自然手写字体。然后显示 ①生成图片、②修改设定、③重新捏人。输入“1”“生成”或“出图”立即根据最终提示词生成；输入“2”后指定项目并重新输出完整设定与提示词；输入“3”清空状态并从①开始。“再来一个”保留 8 项核心设定和固定摄影，只重新演绎动作、环境、时令、天气、构图、镜头和光影，然后生成新的提示词与图片。这是连续状态游戏，不是普通问答。

英語プロンプト

Use the uploaded reference image as the facial identity reference for the main visible adult woman. Run an interactive ancient-beauty character creation game. Keep explanations minimal, save state after every choice and show only one item at a time. Process in order: 1 Dynasty -> 2 Body -> 3 Temperament -> 4 Hairstyle -> 5 Clothing -> 6 Makeup -> 7 Scene -> 8 Expression -> Completion. At the start show only 1 Dynasty. After each choice save immediately and advance without asking whether to continue or repeating confirmation. For items 1-3 provide 6-10 compact visual-card options with number, name and a description of no more than 10 words, in a 2-3 column layout. Dynasty options: Pre-Qin, Qin and Han, Wei-Jin/Southern and Northern Dynasties, Sui and Tang, Five Dynasties and Ten Kingdoms, Song, Liao/Jin/Yuan, Ming, Qing. Body: petite/graceful, slender/soft, natural/proportioned, tall/slender, soft/plump, graceful/elegant. Temperament: elegant, gentle, bright, exquisite, dignified, charming, cold, naive. For 4 provide 6-8 historically grounded female hairstyles consistent with the dynasty. For 5 provide 6-8 female clothing forms grounded in artifacts, paintings, archaeology or literature; prohibit modern new-Chinese, pan-Chinese and fictional film/TV costumes. For 6 provide 6-8 historically grounded makeup styles. For 7 provide 6-8 female life scenes consistent with the dynasty's architecture, lifestyle and social environment. For 8 dynamically generate eight clearly distinct expressions based on all current settings; do not use a fixed expression pool. Linkage rules: 4-7 must match 1. Treat Liao, Jin and Yuan separately. If 1 changes, reprocess only 4-7 while preserving 2-3. If 4-7 changes, check later settings for conflicts and reprocess affected items; always regenerate 8. Operations: Random changes the current unlocked item; Change one replaces its options; Lock prevents Random, Change one and Random All; Modify Setting can explicitly change a locked item; Back returns to the previous item and preserves unaffected state; Random All randomizes only unlocked items from 1 onward; Restart Creation clears all state. Understand direct commands such as changing dynasty, hairstyle or expression. If clicks are unavailable, immediately accept numbers or names. AI may complete movement, gestures, accessories, jewelry, colors, patterns, materials, footwear, props, relationships, composition, lens, lighting, season and weather, but every completion must fit the chosen dynasty and character. Seasons: spring, summer, autumn, winter. Weather: sunny, cloudy, misty, drizzle, snow. Fixed character: adult female, natural body shape, coordinated proportions and realistic skin texture. Fixed photography: mobile lifestyle snapshot with CCD direct flash, native digital-photo rendering and 9:16 vertical format; photography is not a selection item and does not change the historical context. After 8 Expression, output in order: Complete Character Settings; Selected single scene; Final Image Prompt with all eight choices and all AI-completed details; then append the signature ● DeepBlue, where ● is a solid #0B3D91 circle and DeepBlue is white natural handwriting. Then show 1 Generate Image, 2 Modify Settings, 3 Restart Creation. Command 1 or Generate immediately generates from the final prompt; command 2 reopens a selected item and then reprints the complete settings and prompt; command 3 clears state. “Again” keeps the eight core settings and fixed photography but reinterprets only action, environment, season, weather, composition, lens and lighting, then generates a new prompt and image. This is a continuous stateful game, not ordinary Q&A.

API モデルを見る

画像・動画プロンプトをもっと見る · 画像を生成 · 動画を生成 · AI キャンバスで作成