Lemo-Opuscar
39 film styles, each with a short film made entirely in code. 39 种影片风格，每种都配一支完全用代码做出来的短片。
Pick a style, bring your own story, and let your coding agent direct the film. 选一个风格，带上你自己的故事，让你的编程 agent 来当导演。
👋 About me · 关于我
I'm Lemomo (@lemomo-ai). More about me on my profile. 我是 Lemomo，更多信息见我的 GitHub 主页。
Not an awesome list. Every film here was made by me, with Claude Opus 5.5. The styles are tuned for Opus 5.5; other models may not reproduce them.
这不是一个 awesome 合集。 这里所有的片子都是我自己用 Claude Opus 5.5 做的。风格是按 Opus 5.5 调出来的，换成其他模型不保证能做出同样的效果。
Every film was directed, drawn, scored and mixed by an AI agent writing code: canvas and WebGL pages rendered frame by frame, original music from free sample libraries, local text-to-speech. No video generation, no stock footage.
每一支片子都是 AI agent 写代码导演、作画、配乐、混音的：Canvas 和 WebGL 页面逐帧渲染，用免费采样库写原创配乐，本地 TTS 配音。不用视频生成，也不用素材库画面。
How to use · 怎么用
git clone https://github.com/lemomo-ai/lemo-opuscar.git
cd lemo-opuscar
claude # or Codex, Cursor… any agent that reads AGENTS.md
Then just say what you want · 然后直接说：
Make a 45-second film in the watercolor style about the coffee farm my family runs. Warm female narrator.
用 watercolor 风格做一支 45 秒的片子，讲我家经营的咖啡园，温暖的女声旁白。
The agent reads three guides and works like a small studio · agent 会读三份指南，像一个小工作室一样开工：
| File · 文件 | What it gives the agent · 给 agent 的东西 |
|---|---|
DIRECTOR.md · 中文 |
how to direct: story, sound, rhythm, camera, performance, self-checks怎么导：故事、声音、节奏、镜头、表演、自检 |
TECHNIQUE.md · 中文 |
how to build: frame-by-frame rendering, voice, music, mixing怎么做：逐帧渲染、配音、配乐、混音 |
styles/<style>/STYLE.md |
what the style looks and sounds like, and how our demo was made这个风格长什么样、听起来什么样，以及样片是怎么做的 |
It stops twice for you: once to approve the story and storyboard, once to approve the look. Then it produces the film into films/<name>/. Prefer your own project? Point your agent at those three files.
它会停下来找你两次：一次确认故事和分镜，一次确认画面，之后把成片做到 films/<名字>/。想在自己的项目里做？把这三个文件交给你的 agent 就行。
Good to know · 提前知道: a film takes an agent about 30–60 minutes and a fair amount of tokens. You need Node 20+, Chrome, ffmpeg and Python 3.11+; big assets (voice model, sample libraries) download only when a step needs them. 一支片子 agent 大约要工作 30–60 分钟，token 用量不小。需要 Node 20+、Chrome、ffmpeg 和 Python 3.11+；配音模型、采样库这类大文件用到时才下载。
The styles · 风格
Click a frame for its STYLE.md · 点图片看它的 STYLE.md。