Actually, we had found them, the problem was either all of the halos were white on white (thus impossible to determine whether there actually was a halo) or might have come from the fact that the test scan was using a printout that already had halos.