Re: [APML] Announcement: New PixInsight LE Version 1.0.2.143 Available

["Greg Hartke" <ghartke@adelphia.net>]

Hi, Carlos,

Sorry about that. I'm not sure what I did to send out that fragment but it
was something goofy. Also worth mentioning: I had to reformat this from
"Rich Text" to "Plain Text" to get under the APML 10 kB message size limit.
Maybe the limit is a little too restrictive?

When I did the DBE, I allowed the program to generate the sampling. I used
24 horizontal samples of size 16. As your documentation suggested, I zoomed
in and moved sample boxes as required to avoid stars. That didn't take as
long as you might think. Your comment about the order is interesting. I
confess to having done the DBE several times because I inadvertently
destroyed my work a couple of times. Clever boy. Anyway, I thought the
results generated using the 2nd order fit had a little more secular
variation than I might have expected based on what I've seen when I removed
uneven field illumination in PS. For that reason, I tried the 4th order fit
which seemed to result in a smoother background.

BTW, after I thought about it last night I decided the reason there was less
noise in the result might have been because the program was working at a
high level of precision. I'll do some more experimentation, of course, but I
really liked the result a lot.

Fortunately for me, I have a fast computer with a lot of memory so I can be
pretty brutal in choosing parameters. It wasn't slow enough to bother me in
the computations I did.

In PS I like layers for the ability to do one step at a time and to do the
same thing multiple times with easy comparison of results. It's also easy in
PS to undo anything you've done. That's important to someone like me who is
hardly a star image processor like some of the folks who hang out here.
That's not to say I can't work out a way to do these things in PixInsight,
I'm just not that familiar with it so naturally I prefer PS for some
procedures. I did indeed have the Real Time Preview enabled but I'm still an
extreme neophyte.

I'm not saying PS is better than PI at these tasks. I'm simply used to the
procedures in PS. In fact, I'm going back to redo the image in PI yet again
(I saved the image after fixing the background) to gain more experience with
the tools available. I suspect the handwriting is on the wall and PI will
become an important tool for astrophotography processing so it seems to me
as if it would be a good idea for me to get in on the ground floor before
the real version of PI becomes available.

Obviously there's a lot of cool functionality in PI. Like anything else of
that complexity, it takes time to learn how to use it.

Regards,

Greg


----- Original Message ----- 
From: Carlos Milovic F.
To: Discussion of Film Astrophotography
Sent: Wednesday, March 09, 2005 11:18 PM
Subject: Re: [APML] Announcement: New PixInsight LE Version 1.0.2.143
Available


>Even I tried the new DBE tool in PixInsight. I found it to work pretty well
>on a >fairly straightforward image (h and chi Persei), though I did it at
>4th order with >no assumed symmetry on a 35 mm negative originally scanned
>at 4000 dpi.

Hi Greg

I don't recommend you to use higher orders than the default one... If you
put several samples this can yield to a wild behavior in the interpolation,
or in other words, just too many oscilations in the function. In most cases,
vigneting is very soft and smooth.

In such star fields, without large nebulosities, working with automatic
generated samples is a good choise... becouse you don't have limitations to
where to put them, using the symetries isn't required. In fact, I would say
that if you have the oportunity to use real samples instead of "virtual"
ones, just take it.

>(It might take a while to do these computations on some platforms and it
> >even took several minutes on my computer.)

A good idea is to subsample the output. Since PixelMath rescales
automaticaly any image in the list to fit the target one, and becouse the
background is very smooth, there is no problem with working with 1/2, 1/4 or
even 1/8 (at 4000dpi) downsamples. This speeds a lot the interpolation
function calculation in DBE, wich is by far slower than upsampling.

>I still find myself going back to PS for curves and levels. I couldn't get
>use to >the curves tool in PixInsight

Why? Do you miss something? Have you tried the Real Time Preview?

>and I hated giving up the ability to do layers.

For most of the processing steps you don't need the use of layers.
PixInsight works with a 32bits inner accuracy, so it doens't matter how many
curves, or histogram's adjustements you apply... there are so many avalaible
levels that there are almost no rounding errors.
If you used layers to mask process... here you have a very easy to use mask
capability.
If you want to combine two or more images (be sure that they fit
perfectly...), averaging, by SMI or whatever method you like... you have
PixelMath.

With a little of imagination (and expertise), you'll be able to create quite
complex masks, do the same as Layer Masking, perform unsharp masks, and a
lot more.

> I may try some more because I'm sure PixInsight is very powerful.

Be sure of that. If you have an alternative to register (aling) images or
create mosaics, you'll find that you don't need nothing more for most of
your images. In fact, the only processes I miss in LE, comparing to the
current standard edition, are Color Saturation, Deconvolution and Dynamic
Alignement. And I have been using exclusively PixInsight for all of my
images for more than a year. I use it even with my daytime shots.

Regards,

Carlos Milovic F.

_______________________________________________
Astro-Photo mailing list
Astro-Photo@seds.org
http://seds.org/mailman/listinfo/astro-photo