It has been shown previously that multipinhole single photon emission computed tomography (SPECT) calibration only requires a two-point source measurement, without knowledge of the distance between them, to uniquely determine the camera geometry. In this work, we studied how this method propagates measurement noise into errors on the estimated calibration parameters, which in turn will degrade the reconstruction image quality. Furthermore, the effect on the calibration and reconstruction accuracy of fixing either the inter-point or the inter-pinhole distance during calibration is assessed. As expected, including information about the distance between the point sources improved the robustness of the calibration against measurement noise. In addition, it was found that similar robustness and even a superior spatial resolution can be obtained by fixing the distance(s) between the pinhole apertures instead of fixing those between the point sources. Finally, optimal two-point source configurations were determined for calibrating our newly-designed 7-pinhole SPECT system for limited field of view mouse imaging while fixing the inter-pinhole distances.