Matlab code for the adaptive rigidification project. This code base provides an implementation of the adaptive rigidification technique as seen at Siggraph 2022. Matlab is intuitive which makes the project easy to setup and results quickly reproducable. To test the various scenes featured in the paper, simply follow the instructions below then open an example file and click play in the editor. Alternatively the name of the example can be called in the command prompt which will launch the simulation. The scenes feature heterogenous materials, multiple objects, scripted animations, surface mapping and more!
Note that this is not meant for production, it could be more optimized, and there could be some hidden bugs. Please do create a pull request if you find any.
The paper and a few useful videos are available on my website https://www.alexandremercieraubin.com/Work/papers/AdaptiveRigidificationShells/
required: opengl, visual studio (or any C++ compilers), cmake, matlab, gptoolbox(for 3D) GPtoolbox might require: embree https://github.com/embree/embree libigl https://github.com/libigl/libigl boost 1.48 https://www.boost.org/users/history/version_1_48_0.html to be detected automatically on windows use the path C:\Program Files\boost\boost_* Tip: install visual studio before cmake so its compiler is automatically added to cmake
GpToolBox is already included in the repo as a subrepo. Simply call git submodule update --init --recursive.
cmake is needed to build gptoolbox's mex. You will need the boost library.
I recommend not to build gptoolbox with eltopo if you are on windows.
We depend on GPToolbox for contacts(signed distances) as well as loading some mesh files.
Some matlab addons are needed or recommended such as Deep Learning Toolbox, Computer Vision Toolbox, Matlab Compiler, Signal Processing Toolbox, and Image processing Toolbox. Those can be found and downloaded from matlab's add-on menu.
git submodule update --init --recursive ;
#if the recursive update fails then manually go to lib/gptoolbox/mex and build
cd lib/gptoolbox/mex ;
cmake -S . -B build -DCMAKE_BUILD_TYPE=Release -DWITH_ELTOPO=false ;
cmake --build build --target install --config Release ;
xcopy /F /Y Release ./ ;
Once the project is set up, add the files to path by calling addAdaptiveRigidificationToPath(); in the command prompt.
Followed by mexAdaptiveRigidification(); to build the project's mex code.
addAdaptiveRigidificationToPath();
savepath();
mexAdaptiveRigidification();
By now you should be all set.
Try running examples in 3d/examples. For instance while being in the root folder, you could open 3d/examples/clothDisk.m and click play or simply call clothDisk in the command prompt. Why don't you try creating your own scene and play with it?
Please do cite the Adaptive Rigidification of Discrete Shells, and Adaptive Rigidification of Elastic Solids papers if you are using the ideas or code inspired by this simulator. Citations are important for us to obtain funding, and can be used as a proof that our work matters. We are also interested to hear from the companies that wish to use this fully or partially.
If you get a signed-distance error when running the 3D code, then you probably did not set up the compiled mex from gptoolbox properly.
Feel free to contact me if you find bugs and to send pull requests with fixes. We likewise have a private repository that we maintain with new research. Contact us to gain access to the private repository if you wish to collaborate with us on our newest/most exciting stuff. Please cite us when using this simulator in your own research.
Special thanks to the collaborators Alexander Winter, David IW Levin and Paul Kry who participated in creating this simulator. This simulator is now used by many students of the McGill physics-based animation lab . We wish to keep expanding it over time.
Some 3D models were inspired and modified from the arcsim provided files. Some of the animation files like the dancer MOT script come from there too. http://graphics.berkeley.edu/resources/ARCSim/