video-prompt-builder

---
name: video-prompt-builder
description: Use when turning a creative brief into a paste-ready Seedance 2.0 video prompt — pairs a shot-by-shot effects breakdown with the @-reference multimodal system, timestamp shot lists, and >15s extension chaining.
source: https://github.com/songguoxs/seedance-prompt-skill
generated: 2026-07-02T18:43:47.231Z
category: tool
audience: creators
---

## When to use

- Turning a one-line brief ("trail-shoe brand film, golden hour, ~15s") into a complete shot-by-shot Seedance 2.0 prompt
- Planning a video as a four-part breakdown: shot timeline, master effects inventory, effects density map, and energy arc
- Writing multimodal prompts that reference uploaded images, video, and audio via the @图片1 / @视频1 / @音频1 system
- Producing videos longer than 15s by chaining segments with Seedance's video-extension (视频延长) feature
- Calibrating effect density and signature moments to a target duration and aspect ratio

## Key concepts

### Paste-ready prompt + breakdown

Every output leads with a single block the user pastes into 即梦/Seedance, then explains the craft via a four-section effects breakdown so the choices are legible and tunable.

### Prompt formula

The base pattern is (主体) + (动作序列) + (环境/光影) + (镜头语言) + (风格) — subject, action sequence, environment/lighting, camera language, then style keywords. For 13-15s clips, write it as a timestamp shot list.

### Four-section breakdown

Shot-by-shot effects timeline (1-4s blocks with stacked effects + transition logic), master effects inventory (name · count · shots), effects density map (HIGH/MEDIUM/LOW per ~3-6s), and a three-act energy arc.

### @reference system

Cite assets inline with official names @图片1-9 / @视频1-3 / @音频1-3, each stating its purpose (首帧, 运镜, BGM). Distinguish 参考 (borrow a style/motion) from 编辑 (modify the source). ≤12 files total.

### Creative principles

Contrast drives impact (alternate density), every video needs one signature hero effect, transitions are shots not throwaway cuts, specificity beats vagueness, and the energy must resolve intentionally.

### Extension for >15s

Single pass caps at 15s. Generate segment 1, feed it back as @视频1, then 将@视频1延长Xs with a matching hand-off frame between segments. Extension length is the new portion only.

## API reference

```
Text-to-video: (主体) + (动作序列) + (环境/光影) + (镜头语言) + (风格)
```

Base text-only prompt with no reference assets.

```
镜头跟随黑衣男子快速逃亡，后面一群人在追，镜头转为侧面跟拍，人物惊慌撞倒路边的水果摊爬起来继续逃，人群慌乱的声音。
```

```
Shot block: SHOT N (timestamp) — name • EFFECT • visual • camera • timing • transition
```

The unit of the shot-by-shot timeline — name stacked effects precisely and state how each shot exits into the next.

```
SHOT 7 (00:07-00:08) — Multi-Exposure Clone
• EFFECT: composited time-slice / stroboscopic clone
• 8+ instances of the runner along the track, ghosting from ~30% to full opacity
• Static background plate, runner composited at multiple time positions
• This is the SIGNATURE VISUAL EFFECT
```

```
Timestamp shot list (13-15s): 0-3秒:[...] 4-8秒:[...] 9-12秒:[...] 13-15秒:[...]
```

Precisely control each beat of a 13-15s narrative clip.

```
15秒仙侠高燃战斗，金红暖色调，0-3秒：低角度特写紧握雷纹巨剑，剑刃赤红电光爆闪；4-8秒：环绕摇镜快切旋身挥剑迸射红色冲击波；9-12秒：仰拍拉远慢放跃起劈向魔兵群；13-15秒：缓推特写落地收剑，余波微动。
```

```
Extend >15s: 将@视频1延长Xs + [新增内容的时间戳分镜]
```

Chain segments past the 15s cap, each picking up from the prior segment's final frame.

```
将@视频1延长15秒。0-5秒：接上段剑修持剑画面，暗影魔兽飞扑而来，剑修纵身迎敌；6-10秒：空中激战剑气纵横，魔兽斩成灰烬；11-15秒：落地收剑，金色粒子飘散，镜头缓推特写侧脸，音效渐弱。
```

## Gotchas

- Lead with the paste-ready prompt, then the breakdown — never ship only the analysis.
- Seedance does not support negative prompts; declare unwanted elements (字幕/LOGO/水印) in a closing 禁止 block.
- Photoreal human faces in uploaded images/video are auto-blocked by the platform.
- Use official @ names (@图片1, never @img1) and label what each reference is for so images, video, and characters don't get mixed up.
- When extending, the generation length is the NEW portion only (extend 5s → length 5s), and each segment needs an explicit hand-off frame.
- Single generation caps at 15s; describe the visual result, not the editing-software step ('frame scales inward rapidly', not 'keyframe a scale effect').
- No hype language in the breakdown — write like a director's shot notes; default prompt language to Chinese for the native 即梦 platform.

---
Generated by SkillMake from https://github.com/songguoxs/seedance-prompt-skill on 2026-07-02T18:43:47.231Z.
Verify against source before relying on details.

File: ~/.claude/skills/video-prompt-builder/SKILL.md