Recently discovered ferromagnetism of the layered van der Waals material VI$_3$ attracts much research attention. Despite substantial progress,in the following important aspects no consensus has been reached: (i) a possible deviation of the easy axis from the normal to the VI$_3$ layers, (ii) a possible inequivalence of the V atoms, (iii) the value of the V magnetic moments. The theoretical works differ in the conclusions on the conduction nature of the system,the value and the role of the V orbital moments. To the best of our knowledge there is no theoretical works addressing issues (i) and (ii) and only one work dealing with the reduced value of the V moment. By combining the symmetry arguments with density functional theory (DFT) and DFT+$U$ calculations we have shown that the antidimerization distortion of the crystal structure reported in Phys. Rev. B {\bf 99}, 041402(R) (2019) must lead to the deviation of the easy axis from the normal to the VI$_3$ layers in close correlation with the experimental results. The antidimerization accompanied by the breaking the inversion symmetry leads to the inequivalence of the V atoms. Our DFT+U calculations result in large value 0.8\mu_B$ of the V orbital moments of the V atoms leading to reduced total V moment in agreement with a number of experimental results and with the physical picture suggested in Phys. Rev. B bf 101, 100402(R) (2020). We obtained large intraatomic noncollinearity of the V spin and orbital moments revealing strong competition between effects coursed by the on-site electron correlation, spin-orbit coupling, and interatomic hybridization since pure intraatomic effects lead to collinear spin and orbital moments. Our calculations confirm the experimental results of strong magnetoelastic coupling revealing itself in the strong dependence of the magnetic properties on the distortion of the atomic structure.
