Image Data Collection

Hey folks,

I'm just a beginner, but I'm intrigued with how you guys collect image
data over time. I often hear 20+ hours of data or more implying that a
number of exposures have been taken of the same subject over many days,
months, even years to build image data. I assume that this requires
detailed notes as to exposure times, focal length of the scope being used,
camera position, FOV, or additional optics in the train, i.e filters,
flatteners, focal reducers, etc. In these cases, how critical is the
geo-location of the imaging scope in terms of latitude? Or is this
automatically worked out because the celestial object does not change in
position relative to the celestial sphere? Additionally, what methods are
used to accurately position/orient the rephotographed object in the frame to
ensure it will align during the stacking process with previously
photographed images? Is this even critical given the large CCD areas some
of you guys are using?

The other question I have regards image processing. My stacking programs
are K3CCD and Registax. I usually follow up with additional processing
using Adobe Photoshop v.7 to enhance the image and pull out the detail not
evident in the original, stacked image. I'm having problems in that I'm not
able to control the dynamic range of the image. This leads to burned out
centers of galaxies or bright nebula and loss of detail e.g. M31 and M42.
Back in my film days, I could control such issues by dodging and burning
selected problem areas during the printing process. In CCD photography,
this technique is not practical since the adjustments need to be done during
exposure. So, simply -- How do I keep my centers from burning out? Is this
corrected during the processing stage or is it controlled during exposure?
or both. If so, how?

My current imaging setup uses a Takahashi TSA102 with a flattener/reducer,
Sbig ST2000XM camera all on a Losmandy G-11 mount. I'm aware of the 8bit
limitations with Photoshop v.7 and plan to upgrade to CS2. Also, I save my
images as jpegs prior to post-processing. I realize this is part of the
problem, since jpeg compression destroys a lot of available detail that
would otherwise be saved with a fits file.

I have several books about using PS as an astrophotography post processor,
but they are tedious to work through.

Your comments and suggestions are appreciated.

V/r
Scott Busby

On Sat, 2 Dec 2006 09:54:50 -0500, "reconair"
wrote:

I'm just a beginner, but I'm intrigued with how you guys collect image
data over time. I often hear 20+ hours of data or more implying that a
number of exposures have been taken of the same subject over many days,
months, even years to build image data. I assume that this requires
detailed notes as to exposure times, focal length of the scope being used,
camera position, FOV, or additional optics in the train, i.e filters,
flatteners, focal reducers, etc.

Any reasonable camera control software will place most of this
information in the image file header. However, very little is actually
required to stack images. If the data is multispectral, you do need to
know, of course, which subframes were collected through which filters.


In these cases, how critical is the
geo-location of the imaging scope in terms of latitude? Or is this
automatically worked out because the celestial object does not change in
position relative to the celestial sphere?

That's correct. Topocentric coordinates don't impact deep sky images.


Additionally, what methods are
used to accurately position/orient the rephotographed object in the frame to
ensure it will align during the stacking process with previously
photographed images?

There are many effective methods. I normally solve my plates, meaning
that the star field is matched to catalogs and very accurate astrometric
data is written to the header, which is used by the alignment software
(Maxim) to stack the images. Other algorithms involve semi-automatically
matching two or more stars, or finding pattern correlations across many
stars or extended features. Most of these techniques can also compensate
for changes in rotation and even image scale between frames.


The other question I have regards image processing. My stacking programs
are K3CCD and Registax. I usually follow up with additional processing
using Adobe Photoshop v.7 to enhance the image and pull out the detail not
evident in the original, stacked image. I'm having problems in that I'm not
able to control the dynamic range of the image. This leads to burned out
centers of galaxies or bright nebula and loss of detail e.g. M31 and M42.
Back in my film days, I could control such issues by dodging and burning
selected problem areas during the printing process. In CCD photography,
this technique is not practical since the adjustments need to be done during
exposure. So, simply -- How do I keep my centers from burning out? Is this
corrected during the processing stage or is it controlled during exposure?
or both. If so, how?

Burning/dodging techniques are easier and more effective in PS than in
the darkroom. Personally, however, I consider them "cheating", and never
use them for astronomical image processing. It is important to me that
objects of actual equal brightness appear equally bright in the image.
This relationship is lost if you dodge/burn.

You need to start by making sure your data quality is high. Avoid
saturated pixels, or limit them at the worst to a small number of the
brightest stars. When you process, use a stacking technique that doesn't
produced saturated data. Your camera produces 16-bit values. If you add
these during stacking, you must do so with a program that works in a
32-bit or floating point space. If your stacking software calculates a
mean or median, its output can be 16-bit, but it still needs to do its
calculations in a deeper space. Most modern software should handle this
correctly.

Now that you have good data, you need to apply a non-linear transfer
function. There are different ways of doing this, but since you have
Photoshop, I'd use FITS Liberator (a free import filter) to bring the
stacked image into PS, applying a transfer function (such as log-log) in
the process. Then I'd use curves on adjustment layers to fine adjust
things. Non-linear transfer functions let you selectively brighten dim
areas such as nebulosity without blowing out brighter areas at the same
time.


My current imaging setup uses a Takahashi TSA102 with a flattener/reducer,
Sbig ST2000XM camera all on a Losmandy G-11 mount. I'm aware of the 8bit
limitations with Photoshop v.7 and plan to upgrade to CS2.

You have good equipment. There are no 8-bit limitations with PS 7 (CS).
It fully supports 16-bit data, but you need to make sure you actually
bring your stacked images into PS in 16-bit mode. CS2 has a few nice
features, especially in terms of sharpening and noise filters, but
really provides minimal improvements in functionality for astroimagers.
I use it and like it, but you don't need to feel it's a must-have
upgrade.


Also, I save my
images as jpegs prior to post-processing. I realize this is part of the
problem, since jpeg compression destroys a lot of available detail that
would otherwise be saved with a fits file.

In fact, this most likely is your ONLY problem. JPEGs are essentially
unprocessable. Besides the lossy compression, they are only 8-bits deep.
There is only one place for JPEGs in astronomical imaging: the very last
thing you do when creating an image for display on the Web or transport
by email. That's it.


I have several books about using PS as an astrophotography post processor,
but they are tedious to work through.

Your comments and suggestions are appreciated.

I'd change your initial processing steps. Don't use K3CCD or Registax-
these tools are better suited to working with video. Use a single
application for your calibration (dark frame and flat frame
compensation) and for alignment and stacking. I recommend Maxim for
this, but you already have CCDSoft with your camera, so you could
alternatively use that. Make sure you create all your subframes as FITS
format, and that the output of your calibration and stacking process is
a FITS file. This becomes your master frame, which can now be imported
(with FITS Liberator) into Photoshop (CS or CS2) as a 16-bit image and
suitably processed with curve adjustment layers and other tools. Much of
the complexity in processing that people write about involves color
processing, which it sounds like you aren't doing. However, if I've
misunderstood, and you are doing color, I'd recommend that you stop. As
a beginner, you should only be shooting B&W (single channel images).
When you have experience processing those successfully you can move on
to color if you wish.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

"Chris L Peterson" wrote in message
...


...However, if I've
misunderstood, and you are doing color, I'd recommend that you stop. As
a beginner, you should only be shooting B&W (single channel images).
When you have experience processing those successfully you can move on
to color if you wish.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Yes, I'm only doing monochrome for now. In my film days, my primary work
was black and white fine art photography. As such, I still have an affinity
for black and white in astrophotography. However, as a science value,
accurate color rendition for celestial objects is important and should be
conveyed. I've done some color work with a one shot color camera with mixed
results. I'll continue black and white with the ST until I've mastered
image control. Besides, I think Ha filtration is very appealing in black
and white images.

Thanks for your constructive comments. I'll certainly put them to use.

Scott

On Sat, 2 Dec 2006 11:05:13 -0500, "reconair"
wrote:

Yes, I'm only doing monochrome for now. In my film days, my primary work
was black and white fine art photography. As such, I still have an affinity
for black and white in astrophotography.

Me, too. From an aesthetic standpoint, I've never seen a good color
astroimage with the same impact as a good B&W one.


However, as a science value,
accurate color rendition for celestial objects is important and should be
conveyed.

Yes, there is more information in a color image, so it is best to
collect multispectral data for many scientific goals. I use my filters
for all sorts of work, but not usually for aesthetic imaging.


I've done some color work with a one shot color camera with mixed
results.

One-shot cameras are pretty limiting.


I'll continue black and white with the ST until I've mastered
image control. Besides, I think Ha filtration is very appealing in black
and white images.

Ha can be very striking. It also opens up great imaging opportunities if
you are unable to work under real dark skies.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com