Glancing Angle Deposition (GLAD) is a technique used in Physical Vapor Deposition (PVD) to prepare thin films with specific properties. During the deposition process, a tilt is introduced between the sputtered atoms flux and the normal of the substrate surface. By shadowing effect, this induces tilted nano-columns that affect the properties of the coating. To predict these properties, several existing tools simulate the different steps of the PVD deposition. First, a simulation of the sputtering of atoms from a metallic target is made, followed by the computation of the atoms transportation from the target to the substrate. Finally, the growth of the film is computed. All these simulations use point based representation to represent the deposited atoms. However, such representation is not suited for classical finite elements analysis (FEA). In this paper, a methodology for generating FEA meshes from the points produced by film growth simulation is presented. Two major scientific challenges are overcome. Firstly, how to segment the ”film” point cloud into a collection of individual “columns” and secondly, how to generate the meshes of the columns that are approximately represented by points. The point cloud segmentation is computed through neighbourhood notion. The mesh is obtained as an implicit surface by the marching cubes algorithm and smoothed by a Humphrey’s class Laplacian algorithm. Numerical simulations based on the generated FEA meshes will be conducted using Abaqus FEA software.