We propose a concise mesh reconstruction approach for 3D point clouds, which proceeds by clustering the input points enriched with quadric error metrics. As the generators of the clusters are the optimal 3D points for the sum of their quadric error metrics, they are preferably placed on sharp features.