传统中式茶馆群像电影场景提示词
传统中式茶馆群像电影场景提示词

中文版提示词

电影级真人实拍画面，传统中式茶馆二层大厅，晚清时期（约19世纪末至20世纪初），白天室内场景。主体为多名身着长衫及传统服饰的茶客与侍者，环境包含深色老木梁柱、雕花扶手与木楼梯、灰泥墙面、格栅窗、木制茶桌、陶瓷茶壶茶杯、香炉、壁画、烛台、纸灯笼，以及开放式二层回廊与层叠建筑结构。

摄影构图采用2.40:1宽银幕比例，全景群像与宽幅环境镜头，平视机位略带正面观察感；运用对角线构图、三分法、前中后景分层、楼梯引导线及框中框构图。色彩以低饱和棕褐色、木质琥珀色、灰青色与暗红色为主，形成历史电影调性。光影采用自然主义明暗对比，柔和漫射光结合局部烛光暖光，光源包括高侧窗日光、窗棂透射光、蜡烛火焰、室内油灯与灯笼实景光源，避免数字特效，以实景表演、真实道具和现场火焰为主，辅以轻微景深控制与自然光影合成。整体呈现写实电影摄影质感，细腻胶片颗粒、真实木材与织物纹理，非CG渲染风格。

English Prompt

Cinematic live-action realistic shot of a second-floor hall in a traditional Jiangnan-style teahouse or inn, set in the late Qing dynasty (circa late 19th to early 20th century), daytime interior. The scene features multiple tea drinkers and attendants in traditional long robes and period costumes, surrounded by dark aged wooden beams and pillars, carved railings and a wooden staircase, plaster walls, lattice windows, wooden tea tables, ceramic teapots and cups, incense burners, murals, candle holders, paper lanterns, and an open second-floor corridor with layered architectural structures.

The photographic composition adopts a 2.40:1 widescreen aspect ratio, with wide establishing shots and full-frame environmental views. Camera at eye level with a slight frontal observational angle; employing diagonal composition, rule of thirds, foreground-midground-background layering, staircase leading lines, and frame-within-frame structures. Color palette is restrained to low-saturation brown, woody amber, greyish-green, and dark red to achieve a historical film look. Lighting follows naturalistic chiaroscuro, combining soft diffused daylight with localized warm candlelight; light sources include high side windows, lattice-transmitted sunlight, candle flames, indoor oil lamps, and lanterns. No obvious digital effects—emphasizing practical performance, real props, and on-set flames, with subtle depth-of-field control and natural light integration. The final look is a realistic cinematic photographic texture, fine film grain, authentic wood and fabric textures, all achieved without CGI.

该提示词已在以下工具中验证有效：

  • Midjourney – 建议使用 --ar 2.40:1 --style raw --s 200，搭配 --v 6.0 或更高版本，可添加 --stylize 250 增强氛围。
  • Stable Diffusion – 推荐写实类模型（如 Realistic Vision、Juggernaut XL），配合 ControlNet（Depth 和 Canny）锁定构图与层次，开启高分辨率修复（放大2倍）。
  • DALL-E 3 – 直接输入全文，生成后可裁剪为2.40:1比例，注意引导模型理解群像分布。
  • 文心一格 – 对中式古建与历史场景理解良好，建议选择“写实摄影”风格并提高画质。
  • 通义万相 – 中文语义响应准确，可完整保留群像与道具细节，推荐输出宽幅尺寸（如 1920×800）。
  • Adobe Firefly – 适合后期调整光影与材质，初稿生成后可进一步细化纹理。
  • NovelAI – 若需强化人物服饰与面部细节，可借助该工具进行精细化补充。

使用建议：针对不同工具添加相应负面提示词（如“过度饱和、扁平光照、畸形肢体、模糊纹理”），采样步数推荐50~70步以获得最佳质感平衡。