Large mesh simplification for distributed environments

An algorithm is described to simplify arbitrarily large triangle meshes while leveraging the computing power of modern distributed environments. Our method combines the flexibility of out-of-core (OOC) techniques with the quality of accurate in-core algorithms, while representing a particularly fast approach thanks to the concurrent use of several computers in a network. When compared with existing parallel algorithms, the simplifications produced by our method exhibit a significantly higher accuracy. Furthermore, when our algorithm is run on a single machine, its speed is comparable with state-of-the-art OOC techniques, whereas the use of more machines enables relevant speedups. Noticeably, we observe that the speedup increases as the size of the input mesh grows.