PROMPT RADAR / VIDEO视频叙事

One Girl Three Cities One Incredible Journey From London Rain to Paris L

One Girl Three Cities One Incredible Journey From London Rain to Paris Lights to Moscow Snow Made with seedance 2.5 Prompt: 30-sec realistic phone vlog of the same Korean woman, same face/identity, natural skin, same gol

@noorlewisx 840 热度 1.6 万 浏览
编辑先看

它为什么值得进入灵感库

它最有价值的部分,是把一个模糊想法推进到了可以被模型执行的程度。 把一段描述拆成起始状态、动作变化、镜头响应和结束画面,连续性比形容词更重要。

动作与镜头 Prompt

One Girl Three Cities One Incredible Journey From London Rain to Paris Lights to Moscow Snow Made with seedance 2.5 Prompt: 30-sec realistic phone vlog of the same Korean woman, same face/identity, natural skin, same gold hoops. Hair/outfit change only by city: London low side ponytail + olive trench; Paris high bun + beige trench; Moscow loose waves + cream faux-fur coat. 0:00–0:07 LONDON: Rainy Soho selfie, clear umbrella, fish & chips, hot-chip reaction, Tower Bridge behind. Dialogue: “The rain really won't stop.” 0:07–0:10 BIG BEN: Tilt up, landmark only, 70% slow motion. “Wow, that's Big Ben.” Blue-grey torn-paper transition. 0:10–0:17 PARIS: Montmartre night, crêpe, reaction, Arc de Triomphe. “So this is what Paris feels like at night.” 0:17–0:20 EIFFEL TOWER: Lights turn gold, 70% slow motion. “Wow, the lights just came on.” Navy torn-paper transition. 0:20–0:27 MOSCOW: Snowy Arbat, blini, removes right glove, bite, Bolshoi Theatre. “It's snowing so much.” 0:27–0:30 SAINT BASIL'S: Colorful domes through snow, 70% slow motion. “Well, bye everyone!” Left gloved hand covers lens → white. STYLE: Authentic handheld 16:9 phone footage, 26mm, natural movement, realistic skin/hands, no cinematic look. STRICT: English only. No logos, text, subtitles, captions, watermarks, brands, signs, UI, maps or equipment. Same face throughout; no AI-looking features.
查看原作者内容

不要一次改完所有变量

用 Seedance 复刻时,建议先单独生成稳定首帧,再把动作拆成“开始、变化、收束”三段;每段只保留一个主要运动。

  1. 主体:One Girl Three Cities One Incredible Journey From London Rain to Paris Lights to Moscow Snow Made with seedance 2.5 Prompt: 30-sec realistic phone vlog of the same Korean woman, same face/identity, natural skin, same gol
  2. 视觉方向:实时 AI 视频案例;查看原帖媒体与作者说明
  3. 镜头与构图:按原帖镜头流程执行;查看原帖媒体与工作流
  4. 交付约束:auto,保留原作者方法与归属;复刻时避免水印、低清、畸变、主体漂移和无关文字。
负向约束

保留原作者方法与归属;复刻时避免水印、低清、畸变、主体漂移和无关文字。

模型建议

Seedance / Veo / Kling。先用原始 Prompt 建立基准结果,再根据模型对自然语言、镜头运动或文字排版的能力做局部改写。

来源与编辑原则

保留作者,也保留判断。

本页收录公开案例并提供中文索引、结构化拆解和复刻建议,不冒充原作者作品,也不替代原帖上下文。

作者:Noor · 收录于 2026年8月23日

访问原始来源

沿着这条线继续看