Scaled magnification ﹩{{\rm{log}}}_{10}(A\mu )﹩ as a function of normalized source position β/β_outer for the dominant subimage in the Gaussian (h = 2) and top-hat (h = 8) models. Both models with various lens amplitudes A all exhibit similar functional shape well-described by a polynomial form of ﹩{{\rm{log}}}_{10}(A\mu )=58.25{(\beta /{\beta }_{{\rm{outer}}})}^{4}-182.64{(\beta /{\beta }_{{\rm{outer}}})}^{3}+231.48{(\beta /{\beta }_{{\rm{outer}}})}^{2}-151.79(\beta /{\beta }_{{\rm{outer}}})+55.77﹩ (gray dashed line) away from the caustics.