colmap/colmap
FreeCOLMAP - Structure-from-Motion and Multi-View Stereo
FreeFree tier
About colmap/colmap
COLMAP is a general-purpose Structure-from-Motion (SfM) and Multi-View Stereo (MVS) pipeline with a graphical and command-line interface. It reconstructs ordered and unordered image collections into 3D models, supporting a wide range of features for photogrammetry and computer vision. The software is open-source under the BSD license, with pre-built binaries for Windows, Linux/Unix/BSD, Docker images, Conda packages, and Python bindings (PyCOLMAP). Originally developed by Johannes Schönberger at the University of North Carolina at Chapel Hill, it is actively maintained by a team of contributors.
Key Features
General-purpose Structure-from-Motion (SfM) pipeline
Multi-View Stereo (MVS) reconstruction
Graphical and command-line interface
Reconstruction of ordered and unordered image collections
Cross-platform support (Windows, Linux, macOS, BSD)
Python bindings (PyCOLMAP) for integration
Pre-built binaries and Docker images available
BSD-licensed open-source software
Pros & Cons
Pros
- Open-source with permissive BSD license
- Comprehensive pipeline covering both SfM and MVS
- Easy to install via binaries, Docker, or conda
- Active community and regular updates
- Python bindings enable integration into workflows
- Well-documented with tutorials and academic citations
Cons
- Can be computationally intensive for large datasets
- Requires good quality and overlap in input images
- Steep learning curve for beginners without computer vision background
Best For
3D scene reconstruction from photographsPhotogrammetry and surveyingCultural heritage documentationAutonomous navigation and mappingVisual effects and film productionRobotics and SLAM research
FAQ
What is COLMAP?
COLMAP is a general-purpose Structure-from-Motion (SfM) and Multi-View Stereo (MVS) pipeline for 3D reconstruction from image collections.
How can I install COLMAP?
Pre-built binaries for Windows can be downloaded from GitHub releases. Linux/Unix/BSD users can use repology.org. Docker images are available at hub.docker.com/r/colmap/colmap. Conda packages can be installed via conda-forge. For source builds, see the installation guide.
Does COLMAP have Python bindings?
Yes, Python bindings are available via PyCOLMAP on PyPI (pycolmap). CUDA-enabled wheels are also available (pycolmap-cuda12).
How should I cite COLMAP in academic work?
If you use COLMAP in research, please cite the CVPR 2016 paper 'Structure-from-Motion Revisited' by Schönberger and Frahm, and the MVS paper by Schönberger et al.