Compared accuracy of background estimation. Top panel: histograms of the relative error in flux determination on a simulated image in which no background is present; SEXTRACTOR (which automatically computes and subtracts a local background), A-PHOT including background measurement and A-PHOT without background estimation yield very similar results. Bottom panel: same for a simulated image in which a constant background plus local variations in the background intensity have been added; in this case, A-PHOT without background estimate yields incorrect results (out of scale in this plot), while SEXTRACTOR and A-PHOT with background estimation continue to yield similar results.