Introduction
If you want to turn a set of photographs into a 3D scene with free, open-source code, OpenMVG is the framework for you. It is a free and open-source C++ framework for multiple-view geometry and image-based 3D reconstruction, with an end-to-end structure-from-motion pipeline, and it is used by researchers and developers around the world. There is no payment, no account and no serial number, and the code from the official OpenMVG site is the complete framework.
This guide explains what OpenMVG is, what it can and cannot do, what your PC needs to run it, and how to build it from the official website only. You will also get a first-use walkthrough that takes you from building the tools to a reconstructed scene in a few minutes. By the end, you should be able to build OpenMVG on Windows and reconstruct a 3D scene from photographs.
OpenMVG is best for structure from motion, photogrammetry research, and building 3D scenes from photos in code. Because it is free and legitimate, you do not need a cracked copy or an unknown download site. The official website provides the code, and the free version is a real, permanent tool, not a short trial.

What Is OpenMVG?
OpenMVG is a free 3D reconstruction framework that you build on your Windows PC from source and use to reconstruct scenes from photographs. In simple terms, it is a set of libraries and command-line tools where you feed in overlapping photos, and OpenMVG matches features, estimates camera positions and builds a sparse 3D point cloud: you give it photos, run the pipeline, and get a 3D scene. It also exports scenes to tools such as MVE, OpenMVS and PMVS for dense reconstruction.
What makes OpenMVG useful is its focus and its openness. Because it is built for multiple-view geometry, its structure-from-motion algorithms are accurate and well tested, and because it is free software under the Mozilla Public License 2.0, it is free forever with no licence server and no subscription. It also runs on Windows, Mac and Linux, with Android and iOS listed too. Three honest notes: OpenMVG has no built-in dense reconstruction, so you export to another tool for dense point clouds and textured meshes; it is a library and a set of command-line tools, not a desktop app, so you need CMake, Git and a C++ compiler to build it; and it is a research framework, so it has a learning curve and its documentation is incomplete in places.
Key Features
- Structure from motion: Structure from motion turns overlapping photographs into camera poses and a sparse 3D point cloud.
- Feature matching: Feature matching finds the same points across photos, so cameras can be placed accurately.
- Camera calibration: Camera calibration models lenses and focal lengths, so reconstructions stay true to scale.
- Export to dense tools: Export to dense tools such as MVE, OpenMVS and PMVS turns sparse clouds into dense models.
- End-to-end pipeline: An end-to-end pipeline covers the whole reconstruction, from images to a 3D scene.
- Cross-platform framework: A cross-platform framework runs on Windows, Mac and Linux, with Android and iOS listed too.
- Free under MPL-2.0: OpenMVG is free under the Mozilla Public License 2.0, with the source code public.
- Command-line binaries: Command-line binaries run each pipeline step, so you can script and automate reconstructions.
System Requirements
- Operating System: Windows 10 or Windows 11, 64-bit
- Processor: A modern quad-core processor or better
- RAM: At least 8 GB of RAM for basic use, 16 GB or more for large photo sets
- Storage: Around 2 GB of free disk space for the tools, libraries and program files
- Graphics: Any graphics card, for viewing results; reconstruction runs on the processor
- Other: A stable internet connection for the download, Git, CMake and Visual Studio with C++ support, and a mouse for working in 3D
These are practical, modern Windows guidelines, not a word-for-word copy of an official specification table. The most important point is simple: OpenMVG is designed for normal Windows PCs, and a modest modern computer is enough to start.
How to Download OpenMVG (Official Method)
The only safe place to get OpenMVG is the official website. Do not use copies on random file mirrors or “free download” portals, because those copies may be outdated or bundled with unwanted software. The official site is the source of truth for the current version.
Official download link: https://github.com/openMVG/openMVG
On the official OpenMVG site, open the repository and follow the build instructions to clone the code and build it on Windows. There is nothing to buy and no installer to download, because OpenMVG is built from source with CMake. There is no account, serial number or payment involved at any point. If a site asks you to pay for OpenMVG, or offers a cracked version of it, leave that site and return to the official OpenMVG site.
Step-by-Step Installation
Follow these steps in order. OpenMVG is built from source, so you need Git, CMake and Visual Studio with C++ support first. Take your time with the CMake configure and build steps, because those are the steps that most often cause the build to fail on Windows.
- Install Git, CMake and Visual Studio — Install Git, CMake and Visual Studio with C++ support, because OpenMVG is built from source on Windows.
- Open the official OpenMVG site — Open https://github.com/openMVG/openMVG in your browser.
- Clone the repository — Open a command prompt and run git clone –recursive https://github.com/openMVG/openMVG.git, so the code is saved on your PC.
- Create a build folder — Create a new build folder next to the code, so the build files stay separate from the source.
- Configure the build with CMake — Open CMake, point it at the OpenMVG source folder and the build folder, and configure it for Visual Studio.
- Build the project — Open the generated solution in Visual Studio and build it, so the OpenMVG binaries are created.
- Confirm the binaries exist — Check the build folder for the OpenMVG binaries, such as openMVG_main_SfM, so you know the build worked.
Tip: OpenMVG is built from source, so you need Git, CMake and Visual Studio with C++ support first. If the build fails, check that all three tools are installed and that the clone included the submodules, then try again.
How to Reconstruct Your First Scene with OpenMVG
This walkthrough is a simple first project. It is not a full course, but it will help you see OpenMVG working on a real project and learn its main tools.
- Take overlapping photos — Take a set of overlapping photos of a small object, walking around it so every side is covered.
- Copy the photos into a folder — Copy the photos into one folder, so the pipeline can find them all.
- List the images — Run the OpenMVG image listing tool on the folder, so it records the camera data for each photo.
- Compute the features — Run the feature computation tool, so OpenMVG finds matching points across the photos.
- Run the structure-from-motion pipeline — Run the SfM pipeline, so OpenMVG estimates the camera positions and builds the point cloud.
- View the sparse point cloud — Open the result in a viewer, so you can see the cameras and the sparse 3D points.
- Export the scene — Export the scene to a dense reconstruction tool, so you can turn the sparse cloud into a full model.
Is OpenMVG Free?
OpenMVG is free and open source under the Mozilla Public License 2.0. You clone it from the official OpenMVG site, build it, and use it without paying and without a time limit. There is no paid version of OpenMVG itself. Be aware of three honest limits: it has no built-in dense reconstruction, so you export to another tool for dense point clouds and textured meshes; it is a library and a set of command-line tools, not a desktop app, so you need CMake, Git and a C++ compiler to build it; and it is a research framework, so it has a learning curve and its documentation is incomplete in places.
That free status is important for this guide. All the steps here use the official download only, and they do not involve cracks, activators or licence keys. If you ever see a site offering a paid or cracked version of the free program, treat it as unsafe and return to the official site.
Frequently Asked Questions
Is OpenMVG safe to download? Yes, when you get it from the official website only. The official site is the version made by the developers. Avoid third-party sites that bundle extra software or ask for payment for a free framework.
Does OpenMVG work on my Windows PC? Yes, in most cases. OpenMVG runs on Windows 10 and 11 (64-bit), and the system requirements above cover a normal modern PC. Check the official site for the exact current version if your PC is older.
Is OpenMVG really free, or will it stop working? The free version described in this guide is not a short trial. It stays usable after building. There is no paid edition of OpenMVG itself.
Do I need a powerful PC to run OpenMVG? No, not for basic use. A modern Windows 10 or 11 PC with 4 GB of RAM or more can run it for simple projects. Larger projects with many objects and effects will benefit from more RAM and a faster processor, but you can start and learn on a normal home PC.
Conclusion
OpenMVG is a useful, free addition to a Windows setup because it is a free open-source C++ framework for multiple-view geometry and 3D reconstruction from photographs, and it can be downloaded safely from the official site. Start with a small first project, learn one tool at a time, and build from there.
To get started, clone OpenMVG from the official website here: https://github.com/openMVG/openMVG Build it with CMake, run the structure-from-motion pipeline on your photos, and see your first 3D scene appear. With a little practice, it can become a regular part of your workflow.
Official download link: https://github.com/openMVG/openMVG
Related Guides
- COLMAP for PC: How to Download & Install on Windows
- Meshroom for PC: How to Download & Install on Windows