EranOfek
commented
1 day ago This is simply due to the resolution of the search (which is a kind of binary-search using a template-bank with up to five levels). I am considering some changes.
On Sun, Nov 3, 2024 at 3:22 PM EastEriq @.***> wrote:
Judging from what is seen when computing FWHM of the observed images, which is done with imUtil.psf.fwhm_fromBank(Image, 'HalfSize',1000), the results show recurring values, hinting at a possible nonuniform distribution. For instance the values (rounded at two decimals) 2.81, 3.13, 3.49 are particularly frequent. In PIV parlancy, this phenomenon is called pixel-locking, and is caused by biased estimations of the position of the autocorrelation peak. Are you sure that something similar is not happening here too? For the purpose of a coarse estimation of the seeing or of the goodness of focus we can probably live with it, but are you using the function anywhere else in the pipeline, e.g. in PSF estimations? Are there any risks that the bias is propagated, for instance, to photometry?
— Reply to this email directly, view it on GitHub https://github.com/EranOfek/AstroPack/issues/511, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJUQ4OYMASMK5IOQASLE63Z6YPTHAVCNFSM6AAAAABRCXPRUOVHI2DSMVQWIX3LMV43ASLTON2WKOZSGYZTCMRQGM3DQNA . You are receiving this because you are subscribed to this thread.Message ID: @.***>
Judging from what is seen when computing FWHM of the observed images, which is done with
imUtil.psf.fwhm_fromBank(Image, 'HalfSize',1000), the results show recurring values, hinting at a possible nonuniform distribution. For instance the values (rounded at two decimals) 2.81, 3.13, 3.49 are particularly frequent. In PIV parlancy, this phenomenon is called pixel-locking, and is caused by biased estimations of the subpixel position of the autocorrelation peak. Are you sure that something similar is not happening here too? For the purpose of a coarse estimation of the seeing or of the goodness of focus we can probably live with it, but are you using the function anywhere else in the pipeline, e.g. in PSF estimations? Are there any risks that the bias is propagated, for instance, to photometry?