RE: Astrometry errors

["Albertson, Chris" <CAlbertson@primeadvantage.com>]

Andrew,

Maybe this would help?

As a first pass use a well known PSF fitting program
to create a preliminary star list.  Some of these
programs are pretty good at deblending.  As you know
that is one argument for using PSF fitting over
aperture photometry, untangling blends.  You can always
check by subtracting the star list from the image.
daophot does this automatically in some interactive way.

Now that you have a "petty good" reduction use that as
input to your 30 parameter fit.  Most optimizers if
you start with a good guess will go faster.  Or you
build a composite star list replacing daophot's output
with your 30 parameter result _only_ when it is better. 

With luck you can solve both your blend problem (by
punting off to daophot) and go faster then two
images per day.

> -----Original Message-----
> From: Andrew Bennett [mailto:andrew.bennett@ns.sympatico.ca]
> Sent: Monday, November 06, 2000 7:09 PM
> To: aah@nofs.navy.mil; tass@listserv.wwa.com
> Subject: Re: Astrometry errors
> 
> 
> On Mon, 6 Nov 2000 09:58:49 -0700, <aah@nofs.navy.mil> wrote:
> 
> >Andrew's latest results indicate that he is approaching
> >1/75 of a pixel centroiding.  We average around 1/100
> >pixel here for our parallax observations, on well exposed,
> >well-sampled, round, isolated stars.  He should be
> >congratulated for achieving nearly the same precision

<<DELETED>>