Relative error ﹩\delta p/p﹩—defined as ﹩({p}_{\mathrm{fit}}-{p}_{\mathrm{in}})/{p}_{\mathrm{in}}﹩—for the parameters ﹩{R}_{1/2}﹩ (top) and inclination i (bottom), as a function of seeing, redshift, inclination, and size-to-PSF ratio ﹩{R}_{1/2}/{R}_{\mathrm{PSF}}﹩ (from left to right). The curves with increasing thickness correspond to subsamples with different SB levels from SB ﹩\lt \;2\times {10}^{-17}﹩, ﹩2\times {10}^{-17}\;\lt ﹩ SB ﹩\lt \;5\times {10}^{-17}﹩, ﹩5\times {10}^{-17}\;\lt ﹩ SB ﹩\lt \;1\times {10}^{-16}﹩, SB ﹩\gt \;1\times {10}^{-16}﹩ erg s⁻¹ cm⁻² arcsec⁻², respectively, where the zero point (dotted line) has been offset for clarity purposes. SB is the observed surface brightness within the observed half-light radius ﹩{\mathrm{SB}}_{1/2,\mathrm{obs}}﹩ times the seeing-to-size ratio ﹩{R}_{1/2}/{R}_{\mathrm{PSF}}﹩, as in Figure 6. The data points represent the median, and the size of the error bars represents the standard deviation for each of the subsamples. One sees that the GalPaK^3D algorithm recovers the morphological parameters irrespectively of seeing, redshift, and/or intrinsic size.