跳到正文

正在自动登录…

← 返回项目列表

lemomo-ai/lemo-opuscar

AI辅助编程视频剪辑

这是一个为Claude Code设计的影片风格工具箱,提供多种现成风格,用户可选用风格并让编程代理根据自身故事生成短片。适合希望用代码和AI代理制作风格化短片的创作者。

AI 中文解读 · 根据仓库描述与 README 生成 · 2026/10/22 15:06:32 · 查看原文

值得看JavaScript提示词与 Skills兼有 资源合集
累计 Stars
5,188
当前热度
7~30 天 · 热门
仓库创建时间
打开 GitHub ↗项目网站 ↗

选用指南

AI 根据 README 整理 · 2026/10/22 15:06:32

使用方式

通过Claude Code插件市场安装skill,或克隆仓库后在Claude Code中使用;用户用自然语言描述影片风格和主题。

查看 README 依据
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, text-to-speech narration. No video generation, no stock footage. ## How to use Two ways in; the skill is the easiest. ### Option 1: install the skill (recommended) In your terminal: ```sh claude plugin marketplace add lemomo-ai/lemo-opuscar claude plugin install lemo-opuscar@lemolab ```
Then use it from any folder. On first use it downloads the guides, tools and style prompts to `~/lemo-opuscar`, shared by all your films. Each film's project, from source to finished video, goes in the folder you started from. For other agents, copy [`plugin/skills/lemo-opuscar/`](plugin/skills/lemo-opuscar/) into their skills folder. ### Option 2: clone the repo ```sh git clone https://github.com/lemomo-ai/lemo-opuscar.git cd lemo-opuscar claude ``` Films go into `films/<name>/` inside the repo. ### Then just say what you want
> Make a 45-second film in the **watercolor** style about the coffee farm my family runs. Warm female narrator. > Make a 40-second **LiDAR Survey** film about the old quarry in the hills behind our town, from public laser-scan data. Name the style in English or Chinese; the [style index](styles/README.md) lists them all.
部署要求

需要Node 20+、ffmpeg和Python 3.11+(或uv),首次使用会下载指南和风格提示到~/lemo-opuscar。

查看 README 依据
| [`TECHNIQUE.md`](TECHNIQUE.md) | how to build: frame-by-frame rendering, voice, music, mixing | | `styles/<style>/STYLE.md` | what the style looks and sounds like; the story is yours | ### Before you start - A film takes an agent about 30–60 minutes and a fair amount of tokens. - You need Node 20+, ffmpeg and Python 3.11+ (or [uv](https://docs.astral.sh/uv/)); the agent installs the rest. - Default output 1920×1080, 24 fps; other sizes on request.
Then use it from any folder. On first use it downloads the guides, tools and style prompts to `~/lemo-opuscar`, shared by all your films. Each film's project, from source to finished video, goes in the folder you started from. For other agents, copy [`plugin/skills/lemo-opuscar/`](plugin/skills/lemo-opuscar/) into their skills folder. ### Option 2: clone the repo ```sh git clone https://github.com/lemomo-ai/lemo-opuscar.git cd lemo-opuscar claude ``` Films go into `films/<name>/` inside the repo. ### Then just say what you want
已知限制

风格针对Claude Opus 5.5调优,其他模型可能无法复现;制作一部影片约需30-60分钟和较多token。

查看 README 依据
[**▶ Watch**](https://lemomo-ai.github.io/lemo-opuscar/opuscar98/) · [**All 98 styles**](https://lemomo-ai.github.io/lemo-opuscar/opuscar98/#styles) · [**Download 1080p**](https://github.com/lemomo-ai/lemo-opuscar/releases/download/films/opuscar98.mp4) </div> ## 👋 About me I'm **Lemomo** ([@lemomo-ai](https://github.com/lemomo-ai)). More about me on my profile. > **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. ![All styles](docs/cover.jpg)
| [`TECHNIQUE.md`](TECHNIQUE.md) | how to build: frame-by-frame rendering, voice, music, mixing | | `styles/<style>/STYLE.md` | what the style looks and sounds like; the story is yours | ### Before you start - A film takes an agent about 30–60 minutes and a fair amount of tokens. - You need Node 20+, ffmpeg and Python 3.11+ (or [uv](https://docs.astral.sh/uv/)); the agent installs the rest. - Default output 1920×1080, 24 fps; other sizes on request.

README

正在加载 README…

收录信息

仓库创建时间
2026/9/26 18:10:41
首次达标收录
2026/10/9 11:44:24
统计更新时间
2026/10/22 19:25:29

按创建时长与累计 Stars 判断等级。首次达标收录后,即使未达当前门槛仍保留;累计 Stars 和仓库状态会定期更新。确认只有 README 及辅助文件的仓库会停止公开。

模型评估

值得看加权总分 3.9 / 5

新颖度
4.0
实用价值
3.5
成熟度
4.3
实质含量
3.9

由 Jev 模型根据仓库描述与 README 评估,2026/10/16 20:36:27 生成。评分仅供参考,不是代码、安全或许可证审查;热度等级另按创建时长与累计 Stars 计算。

仓库信息

最近推送
2026/10/22 15:07:05
许可证标识
MIT
Fork 数
562
当前状态
未归档

收录记录

按创建时长与累计 Stars 首次收录:小火
首次发现项目