psf fitting question
santiago wrote on Nov 01, 2009
Hi, i've been doing some of fitting with daophot, and the resulting psf profile is the following profile... it is not like common psf profile, is ok in the wings, but the central region is up-side-down !! any ideas what can be went wrong? i send an attachment of psf profile obtained,
http://www.fisica.edu.uy/~sroland/iraf/psf_profile.png
thanks,
http://www.fisica.edu.uy/~sroland/iraf/psf_profile.png
thanks,
Douglas W. Shields wrote on Nov 01, 2009
I have had a similar issue. One of my PSF's looks like an alien head (See profile pic). The wings are nicely Gaussian but there are deep distortions in the center. These distortions correspond to values well below zero.
I have tried the following diagnostics:
1. Double-check the values of the parameters that are the most critical (according to the IRAF/DAOPHOT manual).
2. Try generating the psf from one star, five stars, or...up to 60 stars.
3. Try generating the psf from stars in different fields.
The distortion looks the same no matter how many stars I use. Using different fields results in differently shaped distortions, but the distortions are always present.
Has anyone else encountered this problem?
I have tried the following diagnostics:
1. Double-check the values of the parameters that are the most critical (according to the IRAF/DAOPHOT manual).
2. Try generating the psf from one star, five stars, or...up to 60 stars.
3. Try generating the psf from stars in different fields.
The distortion looks the same no matter how many stars I use. Using different fields results in differently shaped distortions, but the distortions are always present.
Has anyone else encountered this problem?
AnTaR3s wrote on Nov 01, 2009
Hey
I've done lots of photometry testing lately and when I started doing PSF fitting with DAOPHOT some of these bad analytic PSFs showed up. In my case it turned out, that it simply was a bad choice of PSF stars in the frame. I chose to let DAOPHOT decide which star to use, but this strongly depends on the completeness of your coordinate list, since all double stars would have to be listet. Its best to build your PSF interactively to see which stars give the best result and in the next step you can choose exactly these without doing interactive work.
I tried to automate the process, because I have to do time series photometry in crowded fields. I managed to get some results with that by choosing only certain stars in the frame as PSF candidates, but still you will have ~1% of the total flux of a star as residuals after you subtract the analytical model...even with a good model (which is not good enough for high precision photometry); so I switched to a different program in combination with DAOPHOT. I now wonder how people get good PSF photometry from a several 1000 frames...
Set all parameters right and choose good PSF stars interactively!
The hint with trying it just on one star is also a good one, since you will be able to see if you are completely going down the wrong way.
I've done lots of photometry testing lately and when I started doing PSF fitting with DAOPHOT some of these bad analytic PSFs showed up. In my case it turned out, that it simply was a bad choice of PSF stars in the frame. I chose to let DAOPHOT decide which star to use, but this strongly depends on the completeness of your coordinate list, since all double stars would have to be listet. Its best to build your PSF interactively to see which stars give the best result and in the next step you can choose exactly these without doing interactive work.
I tried to automate the process, because I have to do time series photometry in crowded fields. I managed to get some results with that by choosing only certain stars in the frame as PSF candidates, but still you will have ~1% of the total flux of a star as residuals after you subtract the analytical model...even with a good model (which is not good enough for high precision photometry); so I switched to a different program in combination with DAOPHOT. I now wonder how people get good PSF photometry from a several 1000 frames...
Set all parameters right and choose good PSF stars interactively!
The hint with trying it just on one star is also a good one, since you will be able to see if you are completely going down the wrong way.
AnTaR3s wrote on Nov 01, 2009
Something has crossed my mind today about that:
Have you used the task seepsf on the profile? Because what you get is just the convolved image...
Have you used the task seepsf on the profile? Because what you get is just the convolved image...
Last post on Nov 01, 2009