Until fairly recently, deep surveys have covered small enough areas, or hadsufficiently limited science goals that it was either possible to visuallyinspect the entire survey, or alternatively filter the data for the specifictargets of interest. As we move to larger and larger surveys, we will not havethis luxury. Most science investigations will have to be done via databasequeries because reprocessing the entire data set will be prohibitivelyexpensive. This makes it especially important to improve algorithms fordetecting and characterizing faint structures, removing or quantifying theinfluence of neighbors or overlapping objects, so that the database is rich anddetailed enough to enable broad science investigations. It also increases theimportance of carrying out joint processing on complementary surveys andincorporating the measurements into the same or linked databases. In this talk Iwill outline science investigations for future surveys that will requireadvances to detection, characterization and photometry. I will discuss lessonslearned from panchromatic HST surveys such as CANDELS as well as considerationsfor LSST, JWST and WFIRST.