Recovery tests in F444W: the Sérsic parameter fractional error (fitted value minus the true value, normalized by the true value) versus true magnitude (first column from the left) or SNR (second column from the left). From top to bottom, each row shows the fractional error on flux, half-light radius, Sérsic index, and ellipticity. The accuracy of recovered Sérsic parameters decreases with source brightness or SNR. In F444W, pysersic retrieves Sérsic parameters within fractional errors of 10% for the majority (68%) of the instances at a given magnitude or SNR, down to SNR ∼ 30, which approximately corresponds to m_F444W ∼ 26. There is little (<2%) or no systematic offset in the median fractional errors at SNR > 10 (or m_F444W < 27). The panels on the right show the histogram of Sérsic parameter residuals normalized by pysersic uncertainties at SNR > 10, compared to a unit's Gaussian distribution. Notably, pysersic uncertainties encapsulate the discrepancy between true and fitted values exceptionally well for ellipticity. However, pysersic systematically underestimates the uncertainty for flux, half-light radius, and Sérsic index by a factor of roughly 2 (or 1.5, in the case of the Sérsic index), as indicated by broadening of these distributions relative to the unit's Gaussian distribution.