PROMPT RADAR / VIDEO视频叙事
THE KITSUNE BLADE SEQUENCE — SEEDANCE 2.0 + GPT IMAGE 2 STORYBOARD One f
THE KITSUNE BLADE SEQUENCE — SEEDANCE 2.0 + GPT IMAGE 2 STORYBOARD One fox warrior. A minimal white studio backdrop. Every slash reads clean because nothing in the frame is fighting for attention. Most creators build a c
@NexlowX 418 热度 5.5k 浏览
编辑先看
它为什么值得进入灵感库
这条内容值得留下的,不是表面的风格词,而是它提供了一条可以继续实验的画面路径。 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。
01 / 原始方法
动作与镜头 Prompt
THE KITSUNE BLADE SEQUENCE — SEEDANCE 2.0 + GPT IMAGE 2 STORYBOARD One fox warrior. A minimal white studio backdrop. Every slash reads clean because nothing in the frame is fighting for attention. Most creators build a character like this and then generate the fight scene straight from a single hero image hoping Seedance guesses the choreography. It doesn't - you get a great character standing in a great pose cut after cut, never actually fighting. Here's the workflow 1. Build the character concept and lock it in GPT Image 2 first- golden fox ears, long amber hair, flowing gold-and-black robe, calm amber eyes hiding controlled aggression, proportions and fabric behavior fixed before motion enters the picture 2. Build a separate storyboard per scene instead of one giant grid -stillness gets its own board, the charge gets its own, the strike and the aftermath each get their own. More room per panel for camera notes, and one scene can be revised without touching the rest 3. Strip the environment down to a plain white/grey studio backdrop, and let the sword carry the energy instead - a single glowing gold trail following the blade does more for "power" than any explosion behind her would 4. Merge the boards into one sheet only at the end, purely for review and to fit the whole sequence into a single reference video - actual generation still pulls from each scene's board individually 5. Animate each scene from its own storyboard in Seedance 2.0 as a staging and camera-movement reference only, with likeness anchored to the GPT Image 2 sheet throughout Why this works - Splitting boards by scene means every camera setup gets enough space to actually be planned, not crammed into a tiny panel - The glowing blade trail becomes the character's signature -consistent across every cut because it's designed once, not improvised per shot - Each scene can be regenerated or refined on its own without disturbing the sequence around it - The calm extreme close-up after the burst is what sells "predator at rest" - stillness after violence hits harder than more violence would Made with GPT Image 2 (storyboard) + Seedance 2.0 (animation). Full reference sheet + prompt in the comments below↓
02 / 复刻路径
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。
- 主体:THE KITSUNE BLADE SEQUENCE — SEEDANCE 2.0 + GPT IMAGE 2 STORYBOARD One fox warrior. A minimal white studio backdrop. Every slash reads clean because nothing in the frame is fighting for attention. Most creators build a c
- 视觉方向:实时 AI 视频案例;查看原帖媒体与作者说明
- 镜头与构图:按原帖镜头流程执行;查看原帖媒体与工作流
- 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