
2009.10.26

  I'm looking into psphot as applied to diff images.  For now, I am
  running a comparison of a simtest output warp with the convolved
  version of the same image as generated by ppSub.  

  * we are losing the cores of bright stars, probably to the choices
    in the FITS output 16-bit conversion (pixel-value -> NAN)

  * errors for the convolved image seem to be smaller for the same
    instrumental mags than for the warp (check chip).

  * the psf model seems to be quite poor -- a lot of structure in the
    psf core.

  * (note: not a very good match with the warp image, though better)

  * psfsub images seem to be slightly displaced in X-dir.  perhaps
    because the convolution kernel has a bit of X-dir tilt in the
    peak.  Try with PPSUB ISIS orders set to 0.0 

2009.10.27

  I need some psphot options / variants:

  * forced photometry:

    I should be able to do:
    
    psphotForced -file input.fits -psf psf.fits -src src.list output 
    
    and have it only perform the forced photometry on the sources listed
    in src.list.

  * psf model creation

    I should be able to do:

    psphotMakePSF -file input.fits -src src.list output

    and have it generate a PSF model using the sources listed in
    src.list

