Plots of the multiplicative bias m for simulated images, based on Eq. (69), for two different priors P_s(ϵ_s) (filled data points: correct prior; open data points: uniform prior). Different styles for the data points indicate different image sizes r_h, see key inside figure, while colours vary with galaxy type: GLAM template (TMP; red); exponential (EXP; green); bulge-like (DEV; blue). Data points for r_h = 0.3′′, same galaxy type, and same prior are connected by dotted lines to guide the eye. The prior for galaxy sizes t, amplitudes A, and centroids x₀ is uniform giving rise to noise-dependent prior bias; the constant offset of EXP and DEV is due to underfitting. A square image pixel has the size 0.1′′, which also equals the PSF size. Results for m for a larger PSF size can be found in Table 1.