Above: Adjusting the b channel by equal amounts. By making the curve steeper, you increase the visible differences between colors, increasing the color range of flat images.
Astro-Physics AP 900GTO mount (replaced my Losmandy G11)
SBIG STV autoguider.
I am shooting my astronomy images with a Digital Single Lens Reflex (D-SLR) camera, namely the Canon EOS 20Da. This camera has a modified design especially for astrophotography. I always shoot RAW format images.
I shoot 4 minute exposures at ISO 800, with a 45 second delay between exposures to let the camera cool down a bit. UPDATE: Beginning late 2007 I'm now shooting ISO 1600.
I am using ImagesPlus v. 2.82 to convert the Raw images from the 20Da, and for calibrating my images with a "dark frame" and a "flat frame." I shoot darks while I'm changing targets, just getting them in here and there through the imaging session, and as I'm packing up. I then median combine the darks from the night and use them in my processing. I shoot flats during twilight, shooting about 16-20 and median combining them. (Update 11/2006, I am not currently using bias frames!)
In general, the dark frame is subtracted from each image frame, and the flat frame is calculated in to reduce vignetting. I'm using the IP workflow per the tutorial, but...
I often use Registar to combine images. Registar is easier to use than IP for images with different camera locations, and can handle radically different frames of the sky easily. Registar has a means to calibrate the levels of the combined frames that can work well. I actually just had some images of an area in Scorpius that Registar had some sort of problem with, and went back to use IP for that and some others. (11/2006 update: I believe the problem with Registar was due to noise. I'm back to using Registar for just about all my image calibrations. It's just the best...)
Photoshop rules for my final processing.
I use Curves to manipulate the image.
NR: Aside from "pulling the image out of the muck," additional noise reduction techniques must be applied to just about every image due to the extreme "stretching" of typical astronomy images.
Noise reduction approaches: There several approaches/tools that can be used for noise reduction. Great results and ease of use are my top priorities. While some free tools may be available, I'm willing to pay for tools that meet my needs in this area. I use a couple of plug-ins. Neat Image is my favorite, and it is easiest to use when there is a fairly blank area of sky (no stars, galaxy, nebula, etc.) to select. I also use Grain Surgery in cases where I can't easily use Neat Image.
Aside: The best noise reduction in Photoshop alone (i.e., if you just using what comes with Photoshop) is based on strategic selection of areas to blur, with application the Gaussian blur. You can select a color range, or use the wand tool with a small # tolerance like 4. You can do this differently in each color channel.
Some images have issues with gradients, and I use and recommend the GradientXterminator plugin.
A few more (basic) details regarding what I do in Photoshop is at this link: Basic Image Processing in Photoshop
Also, see the LAB curves technique I'm using, described below. This helps bring out the colors.
I've often thought that DSLR images can seem relatively flat and monochromatic when compare with the film images I've been seeing (and producing) over the years. This is true especially in the reds, where DSLRs generally have a relatively weak deep red channel. Popular astrophotography films have had a strong red response, including Kodak's PJ400, LE 400, and Supra 400 (all out of production), as well as Kodak E200 and Fuji's SHQ 200 (both currently still available). A reason red is so important: Hydrogen Alpha nebula, such as the Pelican above, glows at a wavelength of about 656 nanometers. Any sensor, be it film or a CCD/CMOS chip, that has poor response in this area will be omitting a critical component of the image of the sky where H Alpha is present. My previous DSLR, a Nikon D100, had a weak H-alpha response, along with relatively high noise. The Canon EOS 20Da I'm now using is a dramatic improvement in both these area. This camera has a modified design especially for astrophotography, but still shoots perfectly good daylight images without any special filters. (Good info on H alpha here.)
There are lots of ways to "add color" to a flat DSLR astronomy image. A classic: apply an "S" curve to the image. (Image->Adjustments->Curves, pull up the right/bright part of the curve, and pull down the left/dark part of the curve.) You can use Image->Adjust->Hue/Saturations, but purists shudder at this.
However, using this LAB technique has advantages as enumerated in Dan Margulis' books including "Photoshop Lab Color" and "Professional Photoshop." It's pretty simple.
Convert Image into Lab. Alt-I-M-L gets you there (Image->Mode->Lab).
View the "a" channel. Ctrl-1 gets you there. It will look very strange.
Here's a trick: Hit the ` key to the left of the # 1. This displays the image in color, but with a channel selected for any manipulations.
Bring up the curves function. Now, make the curve steeper, but keep the center point the same. The curve is a straight line. Do the same with the b Channel. See example below.
How far to go is a judgment call. If you'd like the subtly change the overall color balance, you can vary the line so it doesn't exactly cross the center point. Don't go too far with this!
Blur a/b channels? Other things you may wish to do in lab include blurring the a and b channels a bit, or maybe a lot, to help get rid of noise. This can help keep star colors under control (see caveats.)
Adjust the Luminosity channel. Be aware, thought, that pulling up the luminosity curve to better show the dim parts of your image will tend to flatten colors/reduce color saturation. (You can go back to step 4 again, if you want.)
Ctrl-' gets you back to all channels selected
Convert back to RGB. Alt-I-M-R gets you there.
Example: check my M31 page, including the images at the grid at the bottom of the page, showing where I started color-wise, and where I ended up.
Caveats: There are many, but the main caveat is that this best with images that are not already highly colorful. This technique also tends to exaggerate chromatic aberrations around stars, so this may need to be mitigated. (I should note that blurring the a/b channels in lab helps with this as described in step 6.)
Above: Adjusting the b channel by equal amounts. By making the curve steeper, you increase the visible differences between colors, increasing the color range of flat images.
Below Left: notice how flat this DSLR image seems; little color.
Below Right: Some color is starting to show after using the lab technique described here.
 A starting point for the M31 images on this page: Pretty Flat! |  1st Cut 20Da image stack. 1200 pixels wide. Some Photoshop Lab processing to accentuate the color differences by steepening the 'a' and 'b' curves. |
| Basic Image Processing in Photoshop
|
|
Check out my Veil Image in the LAB color space "a" channel at the bottom of this page...
Astronomy Pictures: Dick Locke's Astrophoto Gateway page....
Copyright © 1998-2007, Dick Locke. All Rights Reserved. Last update: 11/07/2007
Contact and Image Use Information hits.