I investigated diff_id 65024, skycell 1383.111 to find out some of the causes of the false Mops Detections. 

The following is a rough count of the different classifications of detections:

||type || number ||
||burntool || 49 ||
|| cosmic ray || 20 || 
|| star wing || 88 || 
|| star core || 37 ||
|| round || 40 ||

"Round" refers to objects which appear generally psf-like in the diff image (round and subtending more than a pixel or 2).  Most, when investigated further, are small cosmic rays which are convolved into something extended, but probably a bit smaller than the convolved PSF.  (Edit EAM to further clarify 'round')

Here are some examples from each of the categories.

= Burntool =
Detection found on the location of burns.  These mostly seem to be cases where a detection is found on a burntool residual artifact, not necessarily a missed burn.  Currently, the locations of the burns are marked in the mask, but ignored by psphot.  There are 4 detections on this example burn:

[[Image(mopsburntoolf.png)]]

= Cosmic ray hit =
These are obviously extended cosmic rays.  There are 2 cosmic rays here, each resulting in a detection.

[[Image(mopscrf.png)]]

= Star wings =
In these cases, the fine structure of the wings of the PSF does not match between the two exposures, eg due to changes in the rotator position.  The flux in the wing structures is detected.  In this example case, for 2 bright stars, the wings are detected (also, a burntool in the far right)

[[Image(mopsstarwingsf.png)]]

= Star cores =
In some cases, the core of the stars do not line up precisely due to astrometric error.  These are the classic 'dipole'.

[[Image(mopsstarcoref.png)]]

= "Round" =
Sometimes these are real (astronomical), most of the time these are actually not real, but instead cosmic rays covering 1 or 2 pixels.  This can be distinguished from the warps, or possibly from the diff-image size parameters (second moments).

[[Image(mopsokf.png)]]