The aim of this work is to investigate how photon noise and errors in the retrieval of solar magnetic parameters from measured Stokes profiles influences the extrapolartion of nonlinear force-free coronal magnetic fields from photospheric vector magnetograms. To do so we use a nonlinear force-free extrapolation code based on an optimization principle. The extrapolation methods has been extensively tested and applied to data from various telescopes. Here we apply the code artificial vector magnetograms obtained from 3-D radiation-MHD simulations. As a reference case we compute the coronal magnetic field from an ideal magnetogram and compare the result with more realistic magnetograms based on simulated Solar Orbiter/VIM-measurements. We investigate the effect of noise, ambiguities, spatial resolution, inversion mechanism, of Stokes profiles etc. We rate the quality of the reconstructed coronal magnetic field qualitatively by magnetic field line plots and quantitatively by a number of comparison metrices, e.g., the vector correlation with the exact solution and how accurate the free magnetic energy is computed. Not surprisingly, the instrument effects and noise influence the quality of the nonlinear force-free coronal magnetic field model. The extrapolations from realistic vector magnetograms show a reasonable agreement with the ideal reconstruction, however, and are in particular significantly better than extrapolations based on line-of-sight magnetograms only. High quality VIM data will thus allow reasonably accurate extrapolations that can serve as the basis for magnetic coupling science through a comparison with observations from EUS and EUI.