Communitygithub.com

ZJU-REAL/Easel

Detect beats in background music and switch images/clips on each beat with zoom/flash effects, creating energetic beat-sync short videos for platforms like TikTok, Instagram Reels, or YouTube Shorts.

What is Easel?

The Easel skill generates beat-synced (卡点) short videos by analyzing background music with librosa's beat tracking algorithm. It detects the tempo and beat times, then splits the music into segments aligned to each beat—every 1, 2, or 4 beats depending on the desired pace. For each segment, a source image or short clip is displayed with a hard cut (no transitions) to preserve the rhythmic impact. The skill supports visual effects: a 'zoom' push-in (default) for a dynamic feel, a 'flash' white flash for an energetic punch, or 'none' for clean cuts. Materials (images/videos) are reused cyclically if there are fewer sources than beat segments. The output is an MP4 video matching the music length, with an optional max duration (e.g., 15s for TikTok). The skill also respects user profiles to suggest aspect ratios and style (fast flash vs. gentle zoom) based on account personality, but always asks for explicit confirmation on aspect ratio before rendering. This tool is available for Claude Code, Cursor, or similar AI coding agents that can execute Python scripts and shell commands.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/ZJU-REAL/Easel/tree/HEAD/skills/openclaw/beat-sync-video

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Music Beat-Sync Video (Beat Mashup)

Use librosa to detect background music beats, hard-cut scene transitions on each beat, and add beat-synced effects to create an energetic "beat-sync" short video. Everything goes through skills/shared/scripts/beatsync.py – do not manually count beats or hand-stitch filters – the script handles beat detection, segment splitting, material looping, and audio-video alignment.

For a soft slideshow with fixed duration per image, see slideshow-video; for theme → AI-generated images + voiceover + final video, see auto-short-video.

Input

FieldRequiredDescription
Background musicYesMusic file that determines the beat grid (stronger rhythm = better effect; ask if not provided)
MaterialsYesA list of images and/or short clips (in order, cycled) or a directory
Aspect ratioYesMust ask and wait for confirmation if the user or upstream task has not specified landscape/portrait (or exact resolution); do not silently infer from platform, profile, or defaults; do not repeat if already clarified
Switch frequencyNoHow many beats between switches (default: every 2 beats)
EffectNozoom push-in (default) / flash white flash / none

Output (outputs/themeName/)

  • Beat-sync final video (*.mp4)
  • Report: number of beat segments, total duration, detected BPM, effect used

Execution Steps

Script path (relative to project root): skills/shared/scripts/beatsync.py (build -h / beats -h).

1. (Optional) Preview beats

python skills/shared/scripts/beatsync.py beats --music <music>
# Prints tempo (BPM) and beat time points, used to decide what --every should be

2. Generate beat-sync video

python skills/shared/scripts/beatsync.py build \
  -i 1.jpg 2.jpg 3.jpg 4.jpg --music <music> \
  -o outputs/themeName/show.mp4 \
  --size 1080x1920 --every 2 --effect zoom
  • Materials = images or short clips, can be fewer than beat segments (automatically cycled).
  • --every 1 switches every beat (fast flash, needs many materials); --every 4 switches every bar (gentle).
  • --effect flash adds a white flash at segment start for more "fire"; --effect none clean hard cut.
  • --max-duration 15 limits output length (e.g., 15s for TikTok).
  • Alternatively --images-dir <directory> to batch-feed materials (sorted by filename).

Profile Awareness

  • With profile: platforms.md is only used to suggest aspect ratios – still requires user confirmation. Beat-sync style (fast flash / gentle zoom) and effects should match account personality (trendy/sports accounts → fast flash, lifestyle/healing accounts → zoom gentle).
  • Without profile: ask landscape/portrait first; default to every 2 beats, zoom effect.

Rules

  1. Clearer rhythm = more accurate beat detection; if there is no obvious beat, the script gracefully degrades to equal-interval switching and reports this in the output summary.
  2. Beat-sync uses hard cuts (no transitions) to preserve the rhythmic impact – this is intentional design.
  3. Materials are automatically cycled if insufficient; when abundant, use --every 1~2 so each segment gets a fresh image.
  4. Output length defaults to music length; use --max-duration to cap for platform requirements.
  5. All outputs go into outputs/themeName/.

Reference Sources

Beat detection uses librosa beat_track (onset strength + dynamic programming to find beat points). Hard-cutting images at beat points is the standard practice in beat mashup editing. The beat → segment → per-segment rendering → audio-video alignment pipeline is encapsulated into a deterministic script to avoid manual beat counting and alignment errors.

Individual skills in this repo

This repo contains 9 individual skills — each has its own dedicated page.

ZJU-REAL/Easel

AI 视频生成:文生视频 / 图生视频 / 数字人首帧驱动。通过可插拔 provider(通义万相 Wan / 火山 Seedance / 快手可灵 / OpenAI 兼容)异步生成视频,用户自备 API key。当用户说 AI 视频生成、文生视频、图生视频、AI 生成视频、AI 短视频、让图片动起来、数字人视频、生成一段视频 时使用。与 video-strategy(选型/策略)、video-editing(剪辑处理)、clipify(切片)区别:本 SKILL 是从 0 用 AI 生成新视频。

ZJU-REAL/Easel

通用音频处理:音频剪辑/裁剪、格式转码(mp3/wav/m4a/aac)、音量归一化、从视频提取音轨、多段拼接、淡入淡出、变速(保音高)。当用户说“剪音频”“裁一段”“转成 mp3”“调音量/响度”“提取音轨/扒音频”“拼接音频”“淡入淡出”“加速/减速音频”“变速不变调”时使用。与 audio-denoise 的区别:audio-denoise 专做降噪,audio-editing 做除降噪外的通用音频操作(也内置 denoise 作为兜底)。

ZJU-REAL/Easel

一句话主题 → 成品短视频:自动串联 文案→配图/AI视频→配音→字幕→BGM→合成,把 Easel 制作层零件编排成一条'一键出片'流水线。**单条视频、口播/资讯向,画面默认逐句配图 + Ken Burns 缓动,需要动态时才逐段图生视频**。当用户说 一键生成视频、自动做短视频、主题生成视频、帮我做条视频、口播视频一条龙、自动出片、短视频一键生成 时使用。**有剧情/角色/对白/反转/多集的短剧改用 short-drama(每镜强制图生视频、不用静态图冒充);只写脚本用 video-script;只生成单个片段用 ai-video-gen。**

ZJU-REAL/Easel

自动字幕 / 语音转字幕:把音频或视频里的语音识别成字幕文件(SRT/ASS/TXT/JSON),可选把字幕烧录进视频。当用户说自动字幕、语音转字幕、视频加字幕、上字幕、转录、听写、字幕文件、生成字幕、烧字幕时使用。

ZJU-REAL/Easel

Easel clipify: long English talk → punchline cuts → optional 16:9→9:16 face-pan/split → opus word-by-word ASS burn. Ships scripts + assets/preview.png. Part of 1.6k★ Easel suite.

ZJU-REAL/Easel

>- 通用图像处理加工:改尺寸/缩放、裁剪、补边适配平台尺寸、格式转换(png/jpg/webp)、 压缩到目标大小、加文字或图片水印、圆角、多图拼接、生成缩略图、读图片信息。 基于 image_ops.py 确定性处理。 使用时机:用户说"改尺寸"、"缩放图片"、"裁剪"、"压缩图片"、"加水印"、"转格式"、 "圆角"、"拼图/多图联排"、"缩略图"、"适配小红书/朋友圈尺寸"、"把图 pad 成 1:1"。 和 card-*(卡片类)的区别:card-* 是"HTML 设计→渲染出图", image-editing 只加工已有图片,不负责视觉设计。

ZJU-REAL/Easel

多角色对话配音:按 cast 和逐行对白为不同角色分配音色与情绪,合成多声线音轨和带角色名字幕。 当用户说“多角色/双人/剧本/对话配音、多人对白、不同角色不同声音、有声剧配音”时使用。 单一公共音色用 tts-voiceover;克隆本人音色用 voice-clone;整部短剧制作由 short-drama 编排。

ZJU-REAL/Easel

Easel video-highlights: CN/livestream highlight extract + stable static vertical reframe (sibling to clipify).

ZJU-REAL/Easel

>- 视频制作策略与工具选型:AI 视频生成模型对比、视频脚本结构设计、制作流程规划,覆盖产品演示/解说/社媒短视频场景。 当用户说"视频怎么做""视频选型""用什么工具做视频""视频制作流程""视频策略""视频系列规划"时使用。 和 video-script 的区别:video-strategy 做内容规划与工具选型,video-script 写具体某条视频的脚本。

Related Skills