This information is for how to build your own copy of the SDK binaries. Most users of the Azure Kinect DK should not need to re-build the SDK, if you just need to use the SDK see usage for more information. If you need to build from source, you have to clone the repository instead of downloading the Zip file to ensure all the submodule dependencies in place.
Compilers:
Architectures:
Toolchain files in cmake/toolchains are mainly used for our CI system but can be used to test specific flavors if needed.
Building on Windows and Linux requires an active internet connection. Part of the build will require downloading dependencies.
The following tools are required to build on Windows:
Also make sure the following components are selected:
CMake. Add CMake to the PATH. Must be version 3.10 or higher. Only necessary if your version of VS2017 is less than 15.3 or you have not installed the “Linux development with C++” workload as part of your VS installation.
ninja. Add ninja to the PATH. Only necessary if your version of VS2017 is less than 15.3 or you have not installed the “Linux development with C++” workload as part of your VS installation.
tar. Note, if you are running Windows 10 version 17063 or later, this comes installed by default.
python3. During the install make sure to add python to path.
The following tools are optional:
Doxygen. Add doxygen to the PATH.
Required for building documentation. To use, pass the CMake parameter -DK4A_BUILD_DOCS=1
Clang-Format. Please download clang
v6.0.0 since that is what we are using to format our code. To invoke, call ninja clangformat
If you are building from a command prompt, it must be a x64 Visual Studio
developer command prompt in order for CMake to find the installed compilers.
We build both 32-bit and 64-bit binaries, but 64-bit binaries are the only
binaries that are tested. (The command prompt should be called something like
x64 Native Tools Command Prompt for VS 2017). Note: call the command line tool
with the option -arch=amd64 for x64 builds i.e VsDevCmd.bat
-arch=amd64
NOTE: You can run scripts/verify-windows.ps1 to verify that your Windows PC is setup properly.
The list of libraries and tools that must be installed is maintained in a Dockerfile used by our CI system. Extract and run the install list from the Dockerfile to ensure your machine has required dependencies.
The depth engine is needed as well. The depth engine (DE) is a closed source
binary shipped with the Linux Debian package. As an example, run apt install
libk4a1.3 to install the Azure Kinect 1.3 and get the depth engine. See
using the depth engine for information about
versioning and adding the Microsoft’s Package Repository to your machine. NOTE This step is not
need for building, but is required running the SDK
Create a folder named “build” in the root of the git repo and cd into that directory.
mkdir build && cd build
Run CMake from that directory. The preferred build is ninja. All other generators are untested.
Release Build:
cmake .. -GNinja
Debug Build:
cmake .. -GNinja -DCMAKE_BUILD_TYPE=Debug
Run the build (ninja).
ninja
Visual Studio 2017 supports opening CMake based projects directly. Use File / Open / CMake … to open the root CMakeLists.txt in the project.
The C# Wrapper is not included in the primary CMake build systems, it is built using the standard Visual Studio MSBuild build system. The C# build depends on files generated by the CMake and Ninja build systems. Therefore the CMake configuration steps generates files that will be consumed by the C# Build and Test systems to find the needed components. Therefore, you need to run the CMake configuration and build before loading the C# solution.
NOTE: Running a configure step again, such as opening the folder in Visual Studio, will overwrite the settings and can break the C# build.