The accuracy of a 3D scanner is defined by the closeness of its 3D point measurements to a scanned 3D structure. Here, one of the most straightforward 3D structures for an estimate is a sphere lying on a planar surface.
By scanning this kind of structure and comparing the measured 3D point cloud with their parametric 3D model, the error for each point in 3D space can be estimated. For a standard LilScan 3D Profiler combined with a Prusa printer, a standard deviation of 0.12mm or 120 microns is achieved. This means, by assuming a Gaussian distribution, 95% of all measurements are inside an error interval of two-sigma or +-0.24mm.