metashape-mcp

概述

Metashape MCP Server直接将AI驱动的自动化集成到Agisoft Metashape Professional中，提供对整个摄影测量流程的全面控制。服务器在15个模块中提供106种工具，涵盖从照片对齐到3D模型导出的所有功能，还提供10种实时项目检查资源和6种引导工作流提示。它在Metashape的Python环境中本机运行，同时支持HTTP和stdio传输选项，可处理快速操作和长时间运行的任务而不会出现超时问题。

试试问 AI

装完之后，这里有 8 个你可以让 AI 做的事：

• create_project

  Create a new Metashape project

  试试:Create a new photogrammetry project试试:Start a new Metashape project for processing drone photos
• import_photos

  Import photos into a Metashape project

  试试:Import these photos from the folder C:/drone_photos试试:Add images to my current project
• align_photos

  Perform structure-from-motion to align camera positions

  试试:Align the photos to create camera positions试试:Run structure-from-motion on my imported images
• build_dense_cloud

  Generate a dense point cloud from aligned photos

  试试:Create a dense point cloud from the aligned photos试试:Generate dense point cloud for my 3D model
• build_mesh

  Generate a 3D mesh from the dense point cloud

  试试:Build a 3D mesh from my dense cloud试试:Create a surface model of the terrain
• build_texture

  Apply texture to the 3D mesh

  试试:Add texture to my 3D model试试:Create a realistic texture for the mesh
• generate_dem

  Generate a Digital Elevation Model from the dense cloud

  试试:Create a DEM from my point cloud试试:Generate elevation data for the survey area
• create_orthomosaic

  Create an orthomosaic from the aligned photos

  试试:Create an orthomosaic of the survey area试试:Generate a bird's-eye view map of the site
• export_model

  Export the 3D model to a specified format

  试试:Export my 3D model as an OBJ file试试:Save the textured model in FBX format
• detect_markers

  detect coded markers in the photos

  试试:Find all the ground control points in my photos试试:Detect coded markers for georeferencing
• set_crs

  Set the coordinate reference system for the project

  试试:Set the coordinate system to EPSG:32633试试:Use UTM zone 33N for georeferencing
• execute_script

  Execute custom Python code in Metashape

  试试:Run this Python script: chunk.matchPhotos()试试:Execute custom processing code

可对比工具

安装

git clone https://github.com/jenkinsm13/metashape-mcp.git

python -c "import sys; print(sys.executable)"
pip install "fastmcp>=2.0.0"

python proxy.py

{
  "mcpServers": {
    "metashape": {
      "command": "python",
      "args": ["path/to/proxy.py"],
      "env": {
        "METASHAPE_MCP_PORT": "8765"
      }
    }
  }
}

FAQ

这与直接使用Metashape的Python API有何不同？

您不需要编写Python代码，而是用自然语言描述您想要的内容，AI会生成并执行适当的Metashape脚本。这使得摄影测量处理对非编程人员同样可用。

我可以在没有Metashape GUI的情况下运行此服务器吗？

可以。服务器支持使用`start_mcp_headless.py`的无头模式，可在没有显示器的远程服务器、虚拟机或CI/CD流水线中进行自动化处理。

如果操作超过60秒会发生什么？

stdio代理没有超时限制，而HTTP传输在60秒后可能超时。这使得像密集点云生成这样的长时间操作可以运行数小时而不会中断。

metashape-mcp 对比