When talking about reverse engineering, it is necessary to focus on the management of point clouds. Generally speaking, every 3D scanner device codifies simple and complex geometries providing different point cloud densities as an output. Point cloud density is usually more correlated with the technical specifications of the device employed rather than with the morphology of the object acquired. This situation is due to the frequent use of structured grids by a large quantity of devices. In order to solve this problem, we therefore need to integrate the classical structured grid acquisition with a smart selective one, which is able to identify different point cloud densities in accordance with the morphological complexity of the object regions acquired. Currently, we can reach the destination in many different ways. Each of them is able to provide different performances depending on the object morphology and the performances of 3D scanner devices. Unfortunately, there does not yet exist one universal approach able to be employed in all cases. For this reason, the present paper aims to propose a first analysis of the available methodologies and parameters, in order to provide final users with some guidelines for supporting their decisions according to the specific application they are facing. Moreover, the developed guidelines have been illustrated and validated by a series of case studies of the proposed method.