GEL is a C++ library of geometry processing tools mostly intended for computer graphics applications. In particular, GEL contains
- a fairly mature half-edge based polygonal mesh data structure,
- a graph data structure,
- voxel grid data structures,
- spatial data structures for triangles and points
Unique functionality includes
- Local Separator Skeletonization
- Inverse skeletonization: Face extrusion quad meshes (FEQ) from graphs.
There are also several features that you might find elsewhere but which are included in GEL and work well.
- Garland-Heckbert simplification of triangle meshes,
- signed distance field (SDF) computation,
- iso-surface polygonization,
- edge flipping-based mesh optimization,
- mesh smoothing (including anisotropic smoothing).
GEL also contains a linear algebra library for small vectors and matrices (2D, 3D, and 4D) as well as tools for visualizing meshes using OpenGL. One could argue that using a library such as Eigen for the former and perhaps PolyScope for that latter would be more appropriate. However, having these things be part of GEL makes the library more self contained.
GEL requires a recent C++ compiler but has very few dependencies. For visualization, OpenGL and GLFW are required, but these dependencies can be omitted if visualization is not needed. This somewhat limits the capabilities of GEL, and most projects that use GEL would also require other libraries (such as a linear algebra library) but by not requiring these to be installed as dependencies of GEL, compilation is simplified and porting to new platforms becomes much easier.
PyGEL is a set of Python bindings for a subset of the features in GEL. In particular, PyGEL covers almost all the mesh features. In addition PyGEL has its own viewer based on OpenGL, and PyGEL can be used from Jupyter notebooks. In this case, it is possible to visualize meshes using a plotly widget. A significant benefit here is that when the notebook is exported to HTML, the 3D view comes along. In fact, this makes PyGEL a useful tool for teaching geometry processing since work can be carried out in a Jupyter notebook and assignments submitted as HTML files. Moreover, PyGEL works in Google colab notebooks.
Some installation instructions below. But for more documentation please see the doc directory. There is a doxygen script for creating a reference manual and a latex file intro.tex which explains the basics. Please doxygen or pdflatex your documentation. A license is also found in the intro document.
PyGEL is on PyPI and can be installed with pip. For most potential users, there is no reason to look further than:
pip install PyGEL3D
The PyPI package is called PyGEL3D and not PyGEL. The library relies on OpenGL. Probably this is already installed, but you may have to do it. On Ubuntu Linux
sudo apt-get install libglu1 libgl1
should suffice. On Google Colab, I recommend having this as the first cell:
!apt-get install libglu1 libgl1
!pip install PyGEL3D
If you need or wish to build GEL/PyGEL from source, please download or clone it from https://github.com/janba/GEL. Below you will find instructions for how to build and create a package for PyGEL that can be installed via pip. If you look around you will find some other build options for Windows, Mac and Linux. Since GEL is primarily developed on Mac OS, the Xcode projects are actively maintained and cover everything. The two last build options are not actively maintained. However, they may prove of some use and hence not removed.
If you are using a unix-like command line, build with
mkdir build; cd build; cmake ..; make -j 8 ; sudo make install; cd ..
You can next issue the command
python setup.py bdist_wheel
Install using something like
pip install dist/PyGEL3D-*.whl
For this to work, you need to have wheel and setuptools installed. Also python
and pip
should be version 3 variants. There is shell script called build_pygel.sh
that automates the tasks above. If you want to build PyGEL with a minimum of fuss, type
sh build_pygel.sh
If you run this script the C++ code is compiled and then the PyGEL3D wheel is created and installed. If the PyGEL3D package is already installed it will be replaced.
Compiling both GEL and PyGEL requires that you have OpenGL installed unless you choose not to compile graphics support which you can do by setting Use_GLGraphics
to OFF
in the CMake file. GLFW is also needed, but CMake fetches GLFW from github and compiles it along with the GEL code. If you compile in some of the other ways (e.g. using XCode, Visual Studio) there is no simple way to avoid the dependency on graphics libraries. Thus, if you need to avoid the OpenGL requirements, CMake is the way to go.
GEL comes with a few demo applications. In addition to the requirements above, several of these also require GLUT to be installed. Going forward, we should remove the GLUT dependency and move to GLFW for the applications.
PyGEL has a module called jupyter_display
which produces graphics suitable for Jupyter notebooks. This module is based on plotly which must then be installed for it to work. You will also need numpy. However, if you use pip, these required libraries will be downloaded automatically when you install PyGEL.