What grid sampling should I use when building a DEM or DSM?
The grid sampling (equivalent to pixel size) is essentially the resolution of the end raster product. When deciding on a grid sampling, one should consider the point spacing of the source point cloud and the ultimate use of the raster product. Users tend to use a 1:1 ratio of point spacing to grid sampling so that a point would theoretically fall within each grid cell (of course LiDAR is irregularly spaced so this does not always happen). Reasons to use a larger grid sampling than the point spacing include file size, amount of available RAM, and geographic area. For each halving of the grid sampling, the file size is increased by 4 times.