In HEC-MetVue, large gridded datasets are handled differently than other datasets. Since large grids contain many points, a less memory intensive approach is used, in comparison to the normal Triangulated Regular Network (TRN). Although large grids are still kept as a TRN, the points are lazily loaded in memory to avoid overloading the memory, and preventing coordinate conversion from slowing the grid processing. If the number of points in a grid exceed the Large Grid Cell Count Threshold Global Setting (see "Large Grid Cell Count Threshold" in Global Settings), the TRN points will only be loaded if absolutely necessary. This has some important performance considerations. Although it reduces memory impacts and speeds up some processing (when point locations are not necessary), loading grids with this method can slow down other processes. For example, if projecting with the Surface Interpolation method (see ProjectTinToGrid), the projection will be much slower since the points need to be created whenever they are used.