We know that we are losing SN which are on top of galaxies due to the
percentage increase cut in the substractions. Our current setting is
high because of star subtraction errors. Since galaxies are more
diffuse that stars, we may be able to lower the percentage increase over
most of all of each large galaxy.

The idea then is to take a galaxy catalog, like the APS, and examine the
candidates located within the limits of the galaxy (as defined by the 
ellipse approximating the galaxy shape and size) which are included as
the percentage increase in lowered. If we get few false detections
compared to the number of new real candidates, then we can expand this
to the regular search.