This code reduces the complexity of a 3D mesh (e.g., downloaded from 3D Warehouse) to a low-polygon approximation of it. The algorithm works by gradually relaxing and resampling a high-polygon mesh, starting from the convex hull. After optimization, the mesh does not contain any object interior and captures only the overall shape of the object, neglecting small details. Finally, the mesh can be resampled at coarse granularity and is fast to render and visualize due to its low complexity. We developed this algorithm to reduce meshes from 3D Warehouse for efficient inverse graphics using LIBRENDER, see Displets: Resolving Stereo Ambiguities using Object Knowledge (CVPR 2015). An illustration is provided in the video below. The source code for this project has been tested on Ubuntu 14.04 and Matlab 2014b and is published under the GNU General Public License. If you find this code useful, we would be happy if you cite us:
@inproceedings{Guney2015CVPR,
  author = {Fatma Güney and Andreas Geiger},
  title = {Displets: Resolving Stereo Ambiguities using Object Knowledge},
  booktitle = {Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2015}
}