DB+Grid: a novel dynamic blocked grid for sparse high-resolution volumes and level sets

We have developed a new data structure for the efficient representation of sparse, time-varying volumetric data discretized on a 3D grid. Our "DB+Grid", so named because it is a <u>D</u>ynamic <u>B</u>locked <u>G</u>rid that shares several characteristics with B+Trees (typically employed in databases and file systems), exploits the spatial coherency of time-varying data to separately and compactly encode data values and grid topology. DB+Grid allows for cache-coherent and fast data access into sparse, 3D grids of very high resolution, exceeding millions of grid points in each dimension. Additionally, DB+Grid is very general, since it imposes neither topology restrictions on the sparsity of the volumetric data nor access patterns when the data are inserted, retrieved or deleted. This is in contrast to most existing sparse volumetric data structures, which either assume static data (i.e. values and topology) or fixed data topology (e.g. manifold surfaces) and require specific access patterns to avoid slow random access. Since DB+Grid is a hierarchical data structure, it also facilitates adaptive grid sampling, and the inherent acceleration structure leads to fast algorithms that are well-suited for simulations. As such, DB+Grid has proven useful for several applications that call for very large, sparse, animated volumes, e.g. level sets and fluid simulations. In this talk we will compare DB+Grid with existing state-of-the-art dynamic sparse data structures and showcase applications from the visual effects industry.