Tomography in the post migrated domain is a widely used inversion method to update velocities in depth seismic migration. It takes the observed errors in the Offset Common Image Gathers (OCIG) and finds the best velocity model to minimize these errors. Migration velocity analysis take advantage of this tool, but it also uses another tools that involve different resolutions, therefore an adequately grid spacing should be selected while updating the velocity in the inversion process. We review and propose different grids spacing in the migration velocity analysis implemented for the Hess velocity model (2D synthetic datasets by Hess corporation), in order to use the best trade-off between good image resolution and computational cost.