Re: Coma, PSF fitting etc

[Andrew Bennett <andrew.bennett@ns.sympatico.ca>]

On Fri, 16 Feb 2001 09:53:06 -0800, "Albertson, Chris"
<CAlbertson@primeadvantage.com> wrote:

>
>It would seem to me that a "fit" once found would
>apply to all images made with the same lens.
Sadly, this was not true for the data set we are
analyzing. I am putting together a summary, or at
least trying to.

>  Or at
>least if the first fit was used as the starting point
>for the next image with the same lens you should converge
>very quickly.  The only thing that should change from
>image to image is focus due to temperature change.
Yes, apart from the irregular large changes. It is 2 or 3
times faster starting from the right answer! The next
version of the program will start each time from the
previous fit. For this run, I was ultra-cautious about
this as the process might get lost and waste a lot of
computing time starting from an incorrect previous fit.
Early versions managed to "converge" very happily to
completely erroneous "fits" ... so I manually updated
the starting values only when a new set appeared to have
stabilized.
However, things seem quite stable now.
>
>If this is true then the two hour compute time is only
>a one time process.
With the cleaned up program and a 1GHz computer, it is
down to a few minutes per image. 4 to 6 CDs a day.
>  If your method is used mainly as
>an image dianostic run time does not mater either.
It certainly diagnoses images ... wait for the
published version!

Andrew Bennett, Avondale Vineyard