Vocaloid-style MV pipeline skill
A skill that takes Claude Code from a song and lyrics to a finished Vocaloid-style lyric video, rendered in a browser.
- 271 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skill runs rendering, ffmpeg and Blender scripts and clones the third-party JIZURA repository.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/vocaloid-style-mv-pipelineDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill runs rendering, ffmpeg and Blender scripts and clones the third-party JIZURA repository.
This is third-party code. Review the repository files before installing.
What it does
The kit consists of the `skills/vocaloid-style-mv/` skill and a project template. The skill walks an agent through the stages: listening to the track, concept, assets, storyboard, build, review and delivery. Every frame is drawn by Canvas2D and WebGL2 in a headless browser, Playwright captures frames and ffmpeg encodes the video. The template ships 24 parameterised scenes, 13 transitions, the JIZURA lyric-motion layer, tempo analysis through `librosa`, lyric timing and Blender scripts for toon 3D. The result is a 1920×1080 master, a 1080×1920 vertical version and share encodes. For other agents such as Codex the skill works as plain Markdown with scripts.
Who it is for. Musicians and video makers who want to build lyric videos as code and hand repeatable rendering to an agent.
Good fit when
- A track and lyrics exist and a clip in the V-home handwritten style is needed with frames locked to the beat.
- A horizontal clip must be rebuilt as a vertical one with the text re-laid out rather than cropped.
- The agent must redraw a faulty scene and re-render a four-minute clip in minutes.
- A consistent set of poses for an original character is required through reference images.
Not a fit when
- Classic editing in After Effects or an editor timeline is needed: the project is entirely code.
- The computer lacks Node.js, Python, ffmpeg and Chrome or Edge: rendering will not start without them.
- The voice or likeness of real performers is wanted: the tool is built for your own original works.
Example request
Make a lyric video for song.wav with lyrics.txt in Vocaloid style: check the environment, propose a concept and storyboard and wait for my reply.Limitations
Node.js 20, Python 3.10, ffmpeg with ffprobe, git and Chrome or Edge are required. The author tested platform is Windows 11, while macOS and Linux are supported through the same scripts. Without an NVIDIA GPU encoding falls back to x264 and software WebGL is about ten times slower. Blender 5.x is needed only for toon 3D, otherwise the agent fakes it in 2.5D. Character images come from a separate model, and the skill only describes the method.
How to disable. Delete the `~/.claude/skills/vocaloid-style-mv` folder. Clip projects copied from the template live in your own directories and are deleted by hand.
FAQ
How do I check my computer before starting?
Run `python skills/vocaloid-style-mv/scripts/check_env.py`: it checks Node.js, Python, ffmpeg, a browser, git and the optional components.
Why is the clip treated as code?
Every frame is a pure function of song time. Re-rendering gives the same result, parallel workers can take any frame and segment seams are invisible.
Where do lyric timings come from?
Official lyrics come first. `faster-whisper`, vocal separation and forced alignment supply line timing or cover the case without text. The output is an LRC file with JIZURA markup.
Does the skill work outside Claude Code?
Yes, it was written for Claude Code but consists of Markdown and scripts, so Codex and other agents can follow it.
How long does rendering take?
A 3:46 1080p30 clip of 6,796 frames took about four minutes on an RTX 3080 with three workers. Speed depends on the GPU.
Discussion
No questions or reviews yet.
Related
SkillsThe Video by Reference skill learns style from a sample clip and builds an animated short from code with an original soundtrack
Four skills turn a song with lyrics into a watercolour animated music video: storyboard, parallel painters and render.
A skill that makes a music video for your song: every frame is drawn by code and six steps with pauses lead from analysis to edit
A Claude Code skill that records a product in the browser, composes a video with chapters and captions and masters the MP4 locally
The video-explainer skill turns a script into a vertical 9:16 explainer: narration, captions, Remotion and a pre-render review
A set of SKILL.md skills for short-video and content production: topic, script, voice, avatar, editing, covers and motion graphics
README in short
The project holds a skill with method references, an engine template and a showcase clip with a breakdown of decisions. It describes rendering features, beat locking, lyric handling and delivery of files. MIT license, third-party component details in `THIRD_PARTY_NOTICES.md`.