PSFEX -- psf -- ERROR: Cannot open image

[ana.lalovic]

A bit of everything that bothers...

Dear all,

I have 2 questions. Btw, I am using SExtractor for nearby galaxies that always dominate the image and I need only the segmentation image for now (for gim2d). My goal is to have only one object==galaxy both in the segmentation image and the catalog.

FIRST:
There are two things (this scares me) that give almost the same output:
1) keeping DETECT_THRESH=ANALYSIS_THRESH=2 given DETECT_MINAREA=200
2) enlarging DETECT_THRESH=ANALYSIS_THRESH=3 and lowering DETECT_MINAREA=100
Only in the first case the galaxy in the segmentation image looks more "fuzzy" (having extra bright pixels from the neighborhood). Maybe I lose informations in the second case?

SECOND:
I am an optimist and I have tried to make a PSF with PSFEX, following instructions from README file that said:
$ sex -c myfile.sex image.fits -CATALOG_NAME image.cat
// and put catalog type to FITS_LDAC as instructed
$ psfex image.cat -PSF_NAME image.psf
when I got the warnings:
WARNING: PSF_NAME keyword unknown
WARNING: 1st context group-degree lowered (not enough samples)
However, it creates some catalog.psf image only I cannot open it with ds9, MIDAS, gimp etc. It isn't empty (14 400 B) and the size varies depending on the change of the input parameters in my .sex file. It seems to me it's working but fails at some point. (Of course, for creating a psf image I lowered DETECT_MINAREA to gather stars beside the galaxy).
I have the written output of the program:
$ filename [ext] accepted/total samp. chi2/dof FWHM ellip. resi. asym.
$ 2mass 13/13 0.65 0.60 2.91 0.02 0.01 0.01
Another related issue: when I include more stars (lowering DETECT_THRESH) chi2 improves and FWHM decreases. Is this good? Should I try to lower chi2 as much as possible? Or should I be concerned about FWHM? Is there a single indicator of success? Is this correct: manage parameters (DETECT_TRESH etc.) until APERTURE IMAGE includes all we see in the original image and if PSFEX gives the same estimate for FWHM as IRAF/psf to stop and accept???

[Emmanuel Bertin]

Dear Ana,

(03-04-2010 23:58)ana.lalovic Wrote:  FIRST:
There are two things (this scares me) that give almost the same output:
1) keeping DETECT_THRESH=ANALYSIS_THRESH=2 given DETECT_MINAREA=200
2) enlarging DETECT_THRESH=ANALYSIS_THRESH=3 and lowering DETECT_MINAREA=100
Only in the first case the galaxy in the segmentation image looks more "fuzzy" (having extra bright pixels from the neighborhood). Maybe I lose informations in the second case?

There is nothing "scaring" in having less pixels detected/included in isophotal measurements when the detection/analysis threshold is increased. In addition to loosing some faint detections, you do loose information when using a higher detection threshold than what the background noise allows for, but it affects primarily isophotal measurements such as MAG_ISO. Other measurements, such as MAG_AUTO, XWIN_IMAGE, are only affected to 2nd order.

(03-04-2010 23:58)ana.lalovic Wrote:  SECOND:
I am an optimist and I have tried to make a PSF with PSFEX, following instructions from README file that said:
$ sex -c myfile.sex image.fits -CATALOG_NAME image.cat
// and put catalog type to FITS_LDAC as instructed
$ psfex image.cat -PSF_NAME image.psf
when I got the warnings:
WARNING: PSF_NAME keyword unknown
WARNING: 1st context group-degree lowered (not enough samples)
However, it creates some catalog.psf image only I cannot open it with ds9, MIDAS, gimp etc. It isn't empty (14 400 B) and the size varies depending on the change of the input parameters in my .sex file. It seems to me it's working but fails at some point. (Of course, for creating a psf image I lowered DETECT_MINAREA to gather stars beside the galaxy).
I have the written output of the program:
$ filename [ext] accepted/total samp. chi2/dof FWHM ellip. resi. asym.
$ 2mass 13/13 0.65 0.60 2.91 0.02 0.01 0.01
Another related issue: when I include more stars (lowering DETECT_THRESH) chi2 improves and FWHM decreases. Is this good? Should I try to lower chi2 as much as possible? Or should I be concerned about FWHM? Is there a single indicator of success? Is this correct: manage parameters (DETECT_TRESH etc.) until APERTURE IMAGE includes all we see in the original image and if PSFEX gives the same estimate for FWHM as IRAF/psf to stop and accept???

The PSFEx tool is not stable and officially supported yet so I am afraid I won't be able to help you with this for the moment. Sorry!
- Emmanuel.