Hi All,
I've been working on this script to generate a non-modeled PSF for a while. I had tried scripts like Mike Schuster's PSFEstimator and had manually generated PSFs using DynamicPSF, but for me that left something out. I have a persistent problem where the CCD cover class is slightly tilted with respect to the sensor in my camera (at least, I now believe this is the case, but that's a different story) causing interesting star shapes. These shapes will not be accurately modeled by a Moffat or Guassian function, so when I use a simulated PSF I am not properly describing the actual point spread function of the captured image. So, my idea was to actually sample the stars in the image and stretch them to match the same peak (1.0), background level (0.0) and size (mean star size) then combine them. In my script I used a simple percentile clipping for the stacking process to throw out outlier information like nearby stars or other structures that might be in the window of the star I'm sampling. I also use other information gathered from DynamicPSF to throw out stars that have a peak brightness over 1.0 or aspect ratios or theta's widely different from the average. There is a control for each of both of these on the form as well as the typical controls for star detection (which uses Juan's code to run inline - thanks Juan!).
Please give it a try and let me know what you think, what you would like to see altered, if you have any concerns with the math (which I admittedly fudged, but seem to get really good results from). I would like to make the script more robust and as technically accurate as possible, so any input would be greatly appreciated.
If you dig into the code, try not to laugh too hard at my poor javascript attempt!
Regards,
David
