I did not read the tardis paper, but the mass distribution gives you at least a upper limit, in the sense that if there is not much mass in one direction, then there is also not much deflection. At least in more or less normal models.