For certain diagnostics plots, it is useful to select on specific subsets of the data. An example subset would be the objects that were used actually in the PSF modeling, i.e. those with flag calib_psfUsed=True. When plotting this subset for the coadds of the LSST stack processing of the HSC RC dataset (DM-6816), a large number of patches have no objects for which this flag is set. While the list is not expected to be identical to the list from the visit processing (the flag is propagated based on the fraction of visits overlapping that object and contributing to the coadd that had the flag set for said object), ending up with a list of zero objects used for PSF modeling should be very unlikely. The pattern is quite delineated, with sharp truncation lines beyond which no objects used in PSF modeling are found:

Of note, this does not happen for the same dataset processed through the HSC 4.0.5 stack (DM-9028):