VidMcp

<p align="center"> <img src="https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/07_creator_polish.gif" width="640" alt="VidMCP 1.1 Creator polish" /> </p> <h1 align="center">VidMCP</h1> <p align="center"> <em>智能代理不描述视频，它们剪辑视频。</em> </p> <p align="center"> <a href="https://vidmcp.com"><img alt="官网" src="https://img.shields.io/badge/vidmcp.com-d4ff2a?style=flat-square&labelColor=050507&color=d4ff2a" /></a> <a href="https://pypi.org/project/vidmcp/"><img alt="PyPI" src="https://img.shields.io/badge/PyPI-2affd1?style=flat-square&labelColor=050507" /></a> <a href="https://dhirajlochib.com/"><img alt="作者" src="https://img.shields.io/badge/Dhiraj%20Lochib-b18cff?style=flat-square&labelColor=050507" /></a> <img alt="Python" src="https://img.shields.io/badge/Python_3.11+-ffa24e?style=flat-square&labelColor=050507" /> <img alt="许可证" src="https://img.shields.io/badge/MIT-efefeb?style=flat-square&labelColor=050507" /> </p> <p align="center"> <b>MCP 服务器</b> 用于智能代理视频编辑 — 背景抠图 · 主体背景特效 · 音频优化 · 字幕 · 导出<br/> 由 <a href="https://dhirajlochib.com/">Dhiraj Lochib</a> · <a href="https://vidmcp.com">vidmcp.com</a> · <code>uv tool install vidmcp</code> 构建 </p> --- ## 创作者优化流程 一键发布就绪的谈话头： ```text 方向调整 → 降噪 → 背景音乐 → 字幕 → 可选背景替换 → 导出预设 ``` | 工具 | 功能 | |------|----------------| | `run_talking_head_polish` | 一键完成全部流程 | | `process_audio` | 降噪 · 增益控制 · EQ · 响度归一化 | | `mix_bgm` | 语音下背景音乐（ducking） | | `transcribe_and_caption` | Whisper 语音识别 + 品牌水印 | | `replace_background` | 背景抠图 + 空间 / 模糊 / 实色板 | | `smart_cut_hesitations` | 删除静默 & 填充词 | | `export_render` | `youtube_16x9` · `reels_9x16` · `square_1x1` | | `import_video` | **自动纠正人像旋转** 默认直立 | ```python from vidmcp.tools.creator import run_talking_head_polish print(run_talking_head_polish( "talk.mov", preset="reels_9x16", bg_mode="space", # none | space | blur | solid strength=0.75, bgm_volume=0.35, )) ``` MCP 代理还可以调用 `apply_recipe(..., recipe_name="talking_head_polish")` 或自由文本 `run_intent(...)`。 ### 代理上下文（工具包 + 简洁结果） 当 MCP 主机加载 **~90 个工具模式** 和庞大的 JSON 数据时，代理会感到应接不暇。VidMCP 默认采用小范围的工具包： | 设置 | 默认值 | 目的 | |---------|---------|---------| | `VIDMCP_TOOL_PACK` | `talking_head` | 只暴露特定产品路径的工具 | | `VIDMCP_COMPACT` | `1` | 减少键值对 / 缩短长载荷 | | `VIDMCP_MAX_RESULT_CHARS` | `4000` | 序列化工具结果的硬上限 | **工具包：** `talking_head` · `education` · `vfx` · `admin` · `all` ```text run_intent("为 reels 优化对话头，空间背景，去除填充词", video_path="talk.mov") project_brief(project_id) # 优先于完整的 get_project list_tool_packs() / set_tool_pack("vfx") get_project(id, detail=true) # 仅在需要时获取完整清单 ``` 流程：`talking_head_polish` · `talking_head_reels` · `talking_head_space` · `talking_head_tight` · `talking_head_infographics`. 用于调试的完整表面：`VIDMCP_TOOL_PACK=all`. --- ## 图库 <p align="center"> <img src="https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/07_creator_polish.gif" width="560" alt="Creator 1.1" /><br/> _{<b>1.1 创作者优化</b> — 方向调整 · 降噪 · 背景音乐 · 字幕 · 导出} </p> <table> <tr> <td align="center" width="50%"> <img src="https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/03_behind_subject.gif" width="100%" alt="Behind subject" /><br/> _{<b>主体背景</b>} </td> <td align="center" width="50%"> <img src="https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/02_tesseract.gif" width="100%" alt="Tesseract" /><br/> _{<b>Tesseract</b> · 4D 特效} </td> </tr> <tr> <td align="center" width="50%"> <img src="https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/01_flowfield.gif" width="100%" alt="Flow" /><br/> _{<b>流场</b>} </td> <td align="center" width="50%"> <img src="https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/05_unit_circle.gif" width="100%" alt="Unit circle" /><br/> _{<b>单位圆</b>} </td> </tr> </table> | 示例 | GIF | MP4 | |--------|-----|-----| | **创作者优化** | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/07_creator_polish.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/07_creator_polish.mp4) | | 流场 | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/01_flowfield.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/01_flowfield.mp4) | | Tesseract | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/02_tesseract.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/02_tesseract.mp4) | | 主体背景 | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/03_behind_subject.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/03_behind_subject.mp4) | | 动态 | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/04_kinetic.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/04_kinetic.mp4) | | 单位圆 | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/05_unit_circle.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/05_unit_circle.mp4) | | 流程 | [gif](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/06_pipeline.gif) | [mp4](https://raw.githubusercontent.com/dhirajlochib/VidMcp/main/demos/samples/06_pipeline.mp4) | --- ## 安装 **需求：** Python 3.11+，[ffmpeg](https://ffmpeg.org/)，推荐 [uv](https://docs.astral.sh/uv/) ```bash uv tool install vidmcp vidmcp --doctor # 或 pip install 'vidmcp[creator]' # + faster-whisper + mediapipe ``` ### MCP (Grok / Claude / Cursor) ```bash # Grok grok mcp add vidmcp \ -e VIDMCP_SAM_BACKEND=mock \ -e VIDMCP_WORKSPACE_ROOT=$HOME/vidmcp-workspaces \ -e VIDMCP_TOOL_PACK=talking_head \ -e VIDMCP_COMPACT=1 \ -- uvx vidmcp ``` ```json { "mcpServers": { "vidmcp": { "command": "uvx", "args": ["vidmcp"], "env": { "VIDMCP_SAM_BACKEND": "mock", "VIDMCP_WORKSPACE_ROOT": "/path/to/workspaces", "VIDMCP_TOOL_PACK": "talking_head", "VIDMCP_COMPACT": "1" } } } } ``` 可选的 Apple Silicon 真实背景： ```bash pip install 'vidmcp[mlx]' export VIDMCP_SAM_BACKEND=mlx export VIDMCP_MLX_MODEL_ID=mlx-community/sam3.1-bf16 ``` --- ## 核心产品 | | | |---|---| | **分割** | 文本提示主体分割（SAM / MLX / MediaPipe / mock） | | **合成** | 主体 **后** 的特效 | | **音频** | 降噪，增强，背景音乐 ducking，响度归一化 | | **字幕** | Whisper 时间线 + 品牌水印 | | **教育** | 数学板 + 语音锁定场景 | | **导出** | 16:9 · 9:16 · 1:1 无拉伸 | ```text create_project → import_video (方向调整) → process_audio → mix_bgm → segment_subject | replace_background → transcribe_and_caption → export_render ``` 优先使用高级路径：`run_talking_head_polish` · `apply_recipe` · `run_quality_gated_pipeline`. --- ## 项目布局 ```text src/vidmcp/ MCP 服务器，创作者工具，感知，效果 demos/samples/ GIF · MP4 示例文件 site/ vidmcp.com (静态 Hostinger) configs/ Claude / Cursor 代码片段 ``` --- ## 作者 **Dhiraj Lochib** — 全栈开发，区块链，人工智能 [dhirajlochib.com](https://dhirajlochib.com/) · [GitHub](https://github.com/dhirajlochib) · [LinkedIn](https://www.linkedin.com/in/dhirajlochib-dev/) · dhirajch145@gmail.com --- ## 许可证 [MIT](LICENSE)