PROMPT RADAR / VIDEO视频叙事

I just nailed this video with Seedance 2.5 model Prompt 👇 Use the uploa

I just nailed this video with Seedance 2.5 model Prompt 👇 Use the uploaded reference image as the exact character reference. Preserve her facial identity, eye color, skin tone, hairstyle, makeup, body proportions, and o

@AIwithkhan 512 热度 8.7k 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

I just nailed this video with Seedance 2.5 model Prompt 👇 Use the uploaded reference image as the exact character reference. Preserve her facial identity, eye color, skin tone, hairstyle, makeup, body proportions, and overall appearance throughout the video. She has long black hair in a high ponytail and wears a vibrant orange cropped jacket over a fitted white crop top, a black pleated skirt layered over fitted shorts, white crew socks, chunky sneakers, silver jewelry, and rings. Maintain perfect character consistency in every scene. Create an ultra-realistic cinematic fashion-performance video inside a massive luxury garage transformed into an editorial studio. Include polished concrete floors, a sleek orange sports car, industrial lights, orange and white LED tubes, chrome surfaces, subtle smoke, and dramatic shadows. The video opens with a close-up of her tightening the laces of her sneakers. She stands up, grabs her orange jacket, puts it on, and looks directly into the camera with a confident expression. She walks toward the sports car with effortless swagger, runs her hand across the hood, then leans against it while posing confidently. The camera moves from a low-angle wheel shot into a smooth full-body reveal. She begins dancing beside the car with sharp shoulder movements, quick footwork, controlled spins, and playful hand gestures. The camera rapidly switches between wide shots, close-ups, overhead angles, and low-angle tracking shots. She jumps onto a small performance platform and delivers an energetic choreography sequence while orange lights pulse around her. She spins, steps forward, points toward the camera, then freezes in a powerful fashion pose. The final shot shows her standing in front of the glowing car with her jacket flowing naturally from a subtle fan. She crosses her arms, gives the camera a confident smile, and the camera slowly pulls backward to reveal the entire garage. Style: High-end automotive fashion film, cinematic performance video, luxury editorial aesthetic, dynamic camera movement, realistic reflections, physically accurate fabric simulation, dramatic lighting, shallow depth of field, photorealistic, ultra-detailed, 4K HDR, 24fps, 16:9. Negative Prompt: No subtitles, no text, no watermarks, no duplicate people, no distorted hands, no facial changes, no identity drift, no unrealistic car deformation, no flickering, no cartoon style, no oversaturated colors.
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:I just nailed this video with Seedance 2.5 model Prompt 👇 Use the uploaded reference image as the exact character reference. Preserve her facial identity, eye color, skin tone, hairstyle, makeup, body proportions, and o
  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:Smiling Khan · 收录于 2026年8月11日

访问原始来源

沿着这条线继续看