Right now (as of PR213), background annuli in photometry take a mean of the pixels in the annulus. This will bias high if there are any interloping objects in the annulus. A median woudl be better (but makes estimating the uncertainty harder); another possibility would be sigma-clipping around the median and taking a mean of what's left.
Right now (as of PR213), background annuli in photometry take a mean of the pixels in the annulus. This will bias high if there are any interloping objects in the annulus. A median woudl be better (but makes estimating the uncertainty harder); another possibility would be sigma-clipping around the median and taking a mean of what's left.