CGAL is a C++ library specialized on geometric computations. CGAL is a big project, therefore it is not included natively into SOFA extlibs directory. Here is the way to compile it for SOFA using CMake.

Install the CGAL plugin – Linux

With Linux, follow the next steps to install CGAL:

• Install Boost for use with Sofa. Note that Sofa with CMake requires the compiled libraries,
• Install GMP and MPFR libraries,
• Install CGAL
• With CMake, set the SOFA-PLUGIN_CGALPLUGIN option to On,
• Configure. Set GMP_DIR and MPFR_DIR to the location that you installed GMP and MPFR. Configure again and generate.

You can now compile CGalPlugin.

Install the CGAL plugin – macOS

With macOS, follow the next steps to install CGAL:

• Install Boost, GMP and MPFR libraries with Homebrew

    brew install boost, gmp, mpfr

• Download the sources of CGAL version 4.7 (last stable version for SOFA) here

• Create a folder in your Homebrew directory to build the plugin

    cd /usr/local/Cellar/
    mkdir cgal4.7
    cd cgal4.7

• Build the plugin with Cmake (redirect to the folder you downloaded before)

    cmake ~/Downloads/CGAL-4.7/
    make

• Go to the build directory of your SOFA install and open CMake GUI

    cd ~/path_to_sofa/sofa/build/buildv17.12/
    cmake-gui ../../

     

• Set the PLUGIN_CGALPLUGIN variable and put the path /usr/local/Cellar/cgal4.7/ in the CGAL_DIR in the Cmake GUI

• Create the GMP_DIR variable and set the path /usr/local/Cellar/gmp

• Create the MPFR_DIR variable and set the path /usr/local/Cellar/mpfr

• Set the PLUGIN_IMAGE variable (for the example, see after), configure, generate and compile

Install the CGAL plugin – Windows

Downloads You will need to download the following:

• CGal: This tutorial will describe how to set up CGal using the installer provided by CGal, which is bundled with some additional required libraries.
• Boost: This is required for a number of Sofa plugins

Installing

• Boost: To use the version of Boost you downloaded, set the SOFA-EXTERNAL_BOOST_PATH cmake variable to its directory, for example: SOFA-EXTERNAL_BOOST_PATH = "C:Program Files (x86)boostboost_1_46_1"
• CGAL: Run the installer that you downloaded above. At some point, there will be a dialog box asking about additional libraries that can be installed with your CGAL installation. Be sure that GMP/MPFR is selected.

Compiling CGAL Libraries

• Open the CMake Gui
• Set Where is the source code to the location that you installed CGAL in (example C:Program FilesCGAL-4.1)
• Set Where to build the libraries to the same location
• Press Configure. You shouldn’t have to make any changes to the configuration. Keep pressing Configure until none of the values show up red.
• Press Generate.
• Open the newly created solution file (found in the CGAL directory) in Visual Studio
• If needed, set the Build Configuration to Release. Compile.

Compiling CGalPlugin

• Open the CMake Gui, and point it to your Sofa source and build
• Check the SOFA-PLUGIN_CGALPLUGIN option
• If you haven’t already done so, set SOFA-EXTERNAL_BOOST_PATH to your Boost directory
• Press Configure. Keep pressuring Configure until none of the values show up red.

Note: If you have used the CGAL installer to install GMP and MPFR, it should find the directories for those libraries automatically. If for some reason GMP_DIR and MPFR_DIR are not set, you can direct set them to the appropriate directory. They are both located in your CGal directory under auxiliary/gmp. You can now open your Sofa solution and compile CGalPlugin.

Test your plugin on the examples

You can launch one of the example scenes available from the root (source directory): runSofa applications/plugins/CGALPlugin/scenes/MeshGenerationFromImage.scn This scene takes as input an Image representing a rectangular parallelepiped with 3 different label in the image. The CGAL library allows to mesh the image using different parameters (e.g. element size) depending on the label. The middle zone (blue) has very fine elements, the purple area has medium-size elements and finally the red area includes coarse tetrahedra. The generated mesh can finally be exported.