PROMPT RADAR / VIDEO视频叙事

Made with Seedance 2.0 + GPT image 2 Prompt: Cinematic photorealistic vi

Made with Seedance 2.0 + GPT image 2 Prompt: Cinematic photorealistic viral "epic entrance" video, 15 seconds, single continuous story beat with 4 distinct camera angles, high-energy surreal-but-grounded style, shot for

@itsshara_ai 454 热度 8.2k 浏览
编辑先看

它为什么值得进入灵感库

先别急着照抄整段 Prompt:真正有用的是主体、空间和镜头之间的约束关系。 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

Made with Seedance 2.0 + GPT image 2 Prompt: Cinematic photorealistic viral "epic entrance" video, 15 seconds, single continuous story beat with 4 distinct camera angles, high-energy surreal-but-grounded style, shot for 9:16 vertical. Reference: SUBJECT (@img1), matching his exact face, build, and white suit throughout. (0:00–0:04) — The Run: He sprints confidently across a rooftop at night toward its edge, a dense glowing city skyline behind him, his white three-piece suit and white fedora catching cool ambient city light, dark hair moving with his stride. Camera: a low three-quarter rear tracking shot, positioned just behind and slightly below him, matching his sprint pace, building kinetic energy. Shallow depth of field, cinematic contrast, fine film grain. Audio: driving bass-and-drum build, footsteps on concrete, distant city hum. (0:04–0:06) — The Leap: He reaches the rooftop edge and launches off into open air, arms beginning to spread, the city skyline sprawling dramatically below him. Camera: a wide shot from behind, holding on the suspended mid-air moment just after the jump. Audio: music drops to near silence, wind beginning to rush in. (0:06–0:11) — The Fall: He free-falls facing straight down in a controlled, confident superhero-style pose, arms spread wide, white suit and fedora staying crisp and controlled against the wind (fedora does not blow off), plunging past a dizzying canyon of illuminated skyscraper windows rushing by on both sides. A glowing neon-blue square portal frame grows rapidly larger, centered directly below him. Camera: a vertical top-down POV chase-cam, diving straight down in perfect sync with his fall, creating a vertigo-inducing straight-down perspective the entire beat. Audio: rushing wind, a rising synth swell, a heartbeat-like bass pulse building tension toward the portal. (0:11–0:15) — The Landing: He drops through the glowing neon square portal and lands in a perfect crouched landing pose on a rooftop party space, instantly rising to full height with arms thrown open wide, confident triumphant expression. A crowd of stylish friends surrounding him erupts — cheering, clapping, some recording on phones, several rushing in to hug and swarm him. Camera: a static wide shot at ground level, the neon portal frame centered in the background, holding steady to let the crowd's energy and his landing read clearly with no camera movement — a deliberate stillness payoff after the vertiginous fall. Audio: music hits its full drop, crowd cheering and laughing, celebratory energy, quick overlapping voices. Overall: photorealistic, premium viral-commercial quality, a deliberately different camera angle per beat (low rear tracking → wide suspended shot → vertical POV dive → static landing wide), consistent cinematic color grade and film grain throughout, high contrast night lighting with neon accents, 9:16 vertical, no on-screen text, no watermark, no logos.
查看原作者内容

不要一次改完所有变量

用 GPT Image 2 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:Made with Seedance 2.0 + GPT image 2 Prompt: Cinematic photorealistic viral "epic entrance" video, 15 seconds, single continuous story beat with 4 distinct camera angles, high-energy surreal-but-grounded style, shot for
  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

GPT Image 2 / 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:Shara I Ai Video Creator · 收录于 2026年7月19日

访问原始来源

沿着这条线继续看