
中文版提示词
电影级真人实拍画面,清代宫廷背景(约18至19世纪),白天室内自然光。主体为一位年轻东亚女性贵妇,身着精致刺绣宫装,佩戴珠串项链,头戴旗头,饰有红色玫瑰头饰、金色发簪及翠玉耳坠,面容端庄,凝视前方。环境陈设包含深色木质格栅、半透明窗棂、古典博古架、青花瓷瓶、青铜器、金色器皿、椭圆形落地镜及木质案台,整体为低调的宫廷内宅装饰。
摄影构图采用2.58:1宽银幕比例,近景至中近景肖像,平视正面机位,中轴对称构图,主体居中,浅景深层次将背景虚化。色彩以暖棕、暗红、金色、墨黑与青绿色组成低饱和宫廷调性。光线运用柔和克制的自然写实光,窗透入的条形日光配合室内漫反射补光,形成轻微轮廓分离,背景渐隐。后期强调轻微光学景深、背景虚化、细腻胶片颗粒、自然肤色修饰及柔和高光晕染,无显著数字特效,追求高端写实真人电影质感,类似数字摄影机结合电影级调色与轻度胶片模拟渲染。
English Prompt
Cinematic live-action realistic shot, Qing dynasty court setting (circa 18th–19th century), daytime interior with natural light. The subject is a young East Asian aristocratic woman in elaborate embroidered court attire, wearing a beaded necklace, a traditional Manchurian headdress adorned with red rose ornaments, gold hairpins, and jade earrings, with a dignified gaze looking forward. The environment includes dark wooden lattice screens, translucent window grids, classical display shelves, blue-and-white porcelain vases, bronze vessels, gold utensils, an oval floor mirror, and wooden tables, forming a subdued court interior decor.
Cinematographic composition adopts a 2.58:1 widescreen ratio, close to medium-close portrait, eye-level frontal shot with axial symmetry, centered subject, and shallow depth of field for background layering. Color palette is restrained to warm brown, dark red, gold, ink black, and teal green for a low-saturation court tone. Lighting employs soft, controlled naturalistic illumination—striped daylight through windows combined with indoor diffuse fill—creating subtle edge separation and a gradual background fade. Post-production emphasizes gentle optical depth of field, background blur, fine film grain, natural skin tone refinement, and soft highlight halation, with no obvious digital effects, aiming for a premium live-action cinematic feel reminiscent of digital cinema cameras paired with film-grade color grading and mild film emulation.
该提示词已在以下工具中验证有效:
- Midjourney – 建议使用
--ar 2.58:1 --style raw --s 180,搭配--v 6.0或--v 6.1,可添加--cw 0保持主体居中。 - Stable Diffusion – 推荐写实肖像模型(如 EpicRealism、Realistic Vision),配合 ControlNet(OpenPose 或 Depth)固定姿势与构图,开启高分辨率修复(放大1.5~2倍)。
- DALL-E 3 – 直接输入全文,生成后可裁剪为2.58:1比例,注意强调“正面中轴对称”以避免偏移。
- 文心一格 – 对清代服饰与宫廷元素识别良好,建议选择“写实摄影”风格并提高细节等级。
- 通义万相 – 中文语义理解准确,可完整保留服饰与陈设细节,推荐输出宽幅尺寸(如 2064×800)。
- Adobe Firefly – 适合后期微调光影与肤色,生成初稿后可进行柔和修饰。
- NovelAI – 若需强化面部妆容与发饰细节,可借助该工具进行精细化补充。
使用建议:针对不同工具添加负面提示词(如“不对称、扭曲五官、过度饱和、扁平光”),采样步数推荐50~70步以获得最佳质感与虚化效果。

◯ 评论 0