editing-decision-engine
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseEditing Decision Engine
剪辑决策引擎
Follow shared release-shell rules in:
- release-shell rules
postplus-shared
Use this skill when the user wants a strong short-form edit plan, not just a script rewrite or a generic video breakdown.
This skill is for post-edit decision making across:
- A-roll performance
- B-roll proof footage
- script meaning
- beat timing
- reference-video editing patterns
- final edit packaging
Do not treat this as a video renderer.
First-version goal:
- output a high-quality edit decision package
- not fully automate the final NLE timeline
遵循以下共享发布框架规则:
- 发布框架规则
postplus-shared
当用户需要的是一份完善的短视频剪辑方案,而非仅仅是脚本改写或通用视频拆解时,使用本技能。
本技能适用于以下场景的后期剪辑决策:
- A-roll 表演内容
- B-roll 佐证素材
- 脚本核心含义
- 剪辑节奏把控
- 参考视频的剪辑模式
- 最终剪辑方案打包
请勿将本工具视为视频渲染器。
第一版目标:
- 输出高质量的剪辑决策包
- 不实现最终NLE时间轴的完全自动化
Use For
适用场景
- planning how to cut a talking-head short
- deciding where B-roll should appear
- mapping script beats to visual proof
- deciding when to stay on A-roll vs cut away
- translating a reference video into reusable edit patterns
- packaging a cut plan for Premiere, Final Cut, CapCut, or a human editor
- 规划访谈类短视频的剪辑方式
- 确定B-roll的插入位置
- 将脚本节奏对应到视觉佐证素材
- 决定何时保留A-roll画面、何时切换镜头
- 将参考视频转化为可复用的剪辑模式
- 为Premiere、Final Cut、CapCut或人类剪辑师制定剪辑方案
Trigger Signals
触发信号
Use this skill when the user asks for things like:
- 这条怎么剪
- 根据 A-roll 和 B-roll 设计剪法
- 参考某条视频的剪辑逻辑
- 按脚本和素材做 post-edit plan
- 逐句决定哪里贴屏幕、哪里留脸
- 给我一个时间轴级别的剪辑方案
Do not use this skill when the user only needs:
- raw video analysis without edit decisions
- frame extraction without timeline reasoning
- render generation from image and audio
Route those to:
- a dedicated visual analysis workflow
skills/40-creative/frame-extractionskills/40-creative/video-batch-runner
当用户提出以下类似需求时,使用本技能:
- 这条怎么剪
- 根据 A-roll 和 B-roll 设计剪法
- 参考某条视频的剪辑逻辑
- 按脚本和素材做 post-edit plan
- 逐句决定哪里贴屏幕、哪里留脸
- 给我一个时间轴级别的剪辑方案
当用户仅需要以下服务时,请勿使用本技能:
- 无剪辑决策的原始视频分析
- 无时间轴逻辑的帧提取
- 基于图像和音频生成渲染视频
请将此类需求引导至:
- 专门的视觉分析工作流
skills/40-creative/frame-extractionskills/40-creative/video-batch-runner
Core Principle
核心原则
Do not describe footage only by what appears on screen.
For edit planning, the important question is:
- what does this shot prove
- what beat does it support
- how much attention should it take
- whether it should lead, support, bridge, or punctuate
The same B-roll clip can be:
- primary proof
- support proof
- transition cover
- pace reset
- visual filler
depending on the spoken beat around it.
请勿仅按画面内容描述素材。
对于剪辑规划而言,关键问题是:
- 这个镜头能佐证什么
- 它支撑哪个剪辑段落
- 它需要占据多少注意力
- 它应起到主导、支撑、过渡还是强调作用
同一段B-roll素材可以是:
- 核心佐证素材
- 辅助佐证素材
- 转场覆盖素材
- 节奏重置素材
- 视觉填充素材
具体取决于其对应的旁白段落。
Read These References
参考文档
- workflow and decision rules:
references/workflow.md - core objects and output shapes:
references/schemas.md
- 工作流与决策规则:
references/workflow.md - 核心对象与输出格式:
references/schemas.md
Required Inputs
必填输入
The skill works best when at least these exist:
- script text or spoken transcript
- local A-roll file or a reliable description of the A-roll performance
- local B-roll files or a usable B-roll shot inventory
- intended output length or target platform
Optional but high-value inputs:
- one or more reference videos
- subtitle draft or transcript with timestamps
- previous edit notes
- campaign or persona context
If assets are missing, do not pretend the plan is precise.
Instead:
- state which decisions are grounded
- state which decisions are provisional
- produce the strongest plan possible from available evidence
当具备以下至少几项输入时,本技能的效果最佳:
- 脚本文本或语音转录稿
- 本地A-roll文件或对A-roll表演内容的可靠描述
- 本地B-roll文件或可用的B-roll镜头清单
- 预期输出时长或目标平台
可选但高价值的输入:
- 一个或多个参考视频
- 带时间戳的字幕草稿或转录稿
- 过往剪辑笔记
- 营销活动或人物设定背景
若缺少部分素材,请勿假装方案精准无误。
应采取以下做法:
- 说明哪些决策是有依据的
- 说明哪些决策是暂定的
- 基于现有素材制定最优方案
Workflow
工作流
1. Lock the edit thesis
1. 确定剪辑核心论点
Before building a timeline, identify:
- what the video is trying to prove
- whether it is tutorial-led, viewpoint-led, comparison-led, or proof-led
- what visual contrast drives the piece
Examples:
old fragmented workflow vs in-context workflowtool overload vs cleaner setupannoying to start vs easy to start
在构建时间轴之前,明确:
- 视频想要佐证的核心观点
- 视频类型是教程导向、观点导向、对比导向还是佐证导向
- 驱动视频的视觉反差是什么
示例:
旧的碎片化工作流 vs 场景化工作流工具过载 vs 简洁配置启动繁琐 vs 轻松启动
2. Break the spoken content into beats
2. 将旁白内容拆解为剪辑段落
Do not plan edits sentence by sentence only.
Break into edit beats based on:
- meaning shift
- emotional shift
- proof need
- pace change
Each beat should capture:
- the spoken line or paraphrase
- timing if known
- beat role
- proof requirement
- visual demand
请勿仅逐句规划剪辑。
根据以下维度拆解为剪辑段落:
- 含义转变
- 情绪转变
- 佐证需求
- 节奏变化
每个段落应包含:
- 旁白原文或意译内容
- 已知的时长
- 段落作用
- 佐证需求
- 视觉要求
3. Understand assets semantically
3. 从语义层面理解素材
For A-roll:
- delivery energy
- pauses
- emphasis words
- facial or body moments worth preserving
For B-roll:
- what it literally shows
- what it proves
- where it is strongest
- how long it can stay on screen before feeling repetitive
For references:
- what structural pattern is reusable
- what is surface style only
对于A-roll:
- 表达的能量感
- 停顿节点
- 强调词汇
- 值得保留的面部或肢体动作
对于B-roll:
- 画面实际展示内容
- 能佐证的观点
- 最具表现力的部分
- 重复出现前可停留的时长
对于参考视频:
- 可复用的结构模式
- 仅为表面风格的元素
4. Decide the cut logic
4. 确定剪辑逻辑
For each beat, decide:
- stay on A-roll or cut away
- which B-roll asset to use
- whether the B-roll is primary or supporting proof
- overlay length
- subtitle density
- whether to use punch-in, hold, montage, J-cut, or L-cut
针对每个段落,决定:
- 保留A-roll画面还是切换镜头
- 使用哪段B-roll素材
- B-roll是核心佐证还是辅助佐证
- 叠加时长
- 字幕密度
- 是否使用特写停留、蒙太奇、J-cut或L-cut
5. Package the output
5. 打包输出内容
Default output should include:
- edit thesis
- beat map
- B-roll assignment table
- time-ordered edit decisions
- risks and missing assets
默认输出应包含:
- 剪辑核心论点
- 段落映射表
- B-roll分配表
- 按时间排序的剪辑决策
- 风险点与缺失素材
Release-Shell Execution Contract
发布框架执行约定
- keep edit theses, beat maps, asset notes, and intermediate decision packages
under
<work-folder>/.postplus/editing-decision-engine/ - keep only final user-facing edit plans outside
.postplus/ - start with a bounded first pass on one sequence or one short video before broader edit planning
- if required inputs such as transcript, A-roll context, or B-roll inventory are missing, stop immediately instead of switching to ad hoc shell glue
- 将剪辑核心论点、段落映射表、素材笔记和中间决策包保存至
<work-folder>/.postplus/editing-decision-engine/ - 仅将最终面向用户的剪辑方案保存至 外部
.postplus/ - 在进行更广泛的剪辑规划前,先针对单个片段或单条短视频完成限定范围的首次处理
- 若缺少转录稿、A-roll背景或B-roll清单等必填输入,立即停止操作,切勿临时拼凑替代方案
First-Version Boundary
第一版边界
Keep the first version pragmatic.
Prefer:
- markdown edit plans
- JSON beat maps
- B-roll proof tables
- simple timeline-ready CSV if useful
Do not require in v1:
- direct XML generation for NLEs
- automatic cut rendering
- perfect shot detection
- automatic motion-design systems
第一版需注重实用性。
优先支持:
- Markdown格式的剪辑方案
- JSON格式的段落映射表
- B-roll佐证表
- 若有用则提供简单的可导入时间轴的CSV文件
第一版无需支持:
- 直接为NLE生成XML文件
- 自动剪辑渲染
- 精准镜头检测
- 自动动态设计系统
Output Standard
输出标准
A strong result from this skill should let a human editor start cutting immediately.
Minimum bar:
- they can tell what the first 3 seconds should do
- they know where the proof moments land
- they know which B-roll is essential vs optional
- they know where the edit should breathe instead of over-cutting
If the output cannot guide a real editor, it is not specific enough.
本技能的优质输出应能让人类剪辑师立即开始剪辑。
最低标准:
- 明确前3秒的画面内容
- 明确佐证节点的位置
- 明确必备与可选的B-roll素材
- 明确剪辑中需要留白而非过度切割的位置
若输出无法指导实际剪辑,则说明不够具体。