The first time you shoot images you might feel like you must have done something wrong. I did.
This page is meant as a demo of one way to bring out the details of an astro-photo. I’ve skipped over lots of details so as to not lose sight of the overall process steps. There are certainly better methods for accomplishing these tasks, but this is the workflow I’ve been using lately. I’m always searching for better ways and I’ll try to update this page when I learn better tools and techniques.
If you have comments, suggestions, or corrections, please email me @ at<at>alberts-astro.com
So….. Here goes:
On your first try shooting a deep space object, your very likely going to be dismayed by how terrible the images look. “Holy smokes, I put in all those hours and can hardly see anything. What’s wrong?” Well, probably lots, but you’ll need to do some processing work to to really find out.
This page is intended to cover a few techniques to “develop” the images you’ve captured into a recognizable picture. It won’t cover the topics of obtaining and stacking the raw images. The scope of those subjects is just to broad to cover here (I may try to put together a “basics” page for those subjects at a later date). So, the assumption here is that you have your image freshly stacked from your favorite stacking software (DSS, AIP4Win, whatever) and that it is in FITS or TIF format and at least 16-bit color. The techniques of course apply to 8-bit images, but you’ll find the limited shades of colors annoying.
We’ll do this by example. Showing the original image, focusing on the important pixels by manipulating the histogram (stretching), applying curves to enhance the brightness of the image, and reducing gradients in the image.
The subject image is a rather poor image I took of M78, a reflection nebula in Orion. It has some bright regions, but is surrounded by dark gas clouds…. it’s a very tough target to get good data on from my light polluted back yard :). The program I’m going to use for this example is PixInsight Le (the free version) however, there are several other programs that can be used to accomplish the same tasks. Some programs do some tasks easier/better, but that’s not important for this demo. The intent is simply to show the basic principles of what can be done.
OK… so here is the starting picture (saved after stacking in DSS with no adjustments):
Not much there eh? Well, actually, there’s a lot there, we just can’t see it yet. The first step will be to stretch the histogram out so we can see what we have.
So opening the Histogram tool (in Photoshop I think you’d use Levels) we can see that all the data is contained in a very small portion of the histogram… see the narrow vertical line on the left side of the left histogram…. that’s it!. Again, different programs deal with this in different ways, but what we’re going to so is amplify those pixels in a non-linear way and clip some of the low end off the histogram. What you see in the panes below is that I’m moving the left and middle sliders to focus only on the area where the bulk of the data is. I do the process twice here (2nd and 3rd panes). I’m actually clipping too much off the bottom here in the right hand pane, but heck this is just a demo.
Now this is what the image looks like after applying the histogram changes:
Wow! There is some data there… and some ugly gradient data from light polution as well. We’ll have to do something about that. However, before we go after the gradient, let’s apply some initial curves to the data to make it a bit easier (mainly so we can see it better).
So let’s open the Curves tool. You can see what I’m doing here. I’m darkening the dim information a little with minimal effect or slight brightening of the brighter data. Note that we could take the background down further, but we need to try and preserve the look of the dark gas clouds that are quite widespread across the frame.
Adjusting curves is very subjective, adjust to suit your tastes, but be careful to try and preserve the original image… after all, the goal is to represent reality, not your artistic talents :). Also be careful to avoid taking any background pixel values to zero. Remember, the sky is not black.
In this case, I took the general RGB down a little, the green channel down just a tad (since it was a big component of the gradient), and the Luminance down a tad also. I checked myself, by looking at the average pixel values in the upper left background to be sure the R, G, & B values were roughly equal.
Ok… So let’s go after the gradient. This can be done different ways (again different programs have different methods) but essentially what we do is make a map of the gradient and subtract it from the original. You might have to do it a couple of times on any given image depending on how good your map is and the software routine used. We’ll just do it once using the Dynamic Background Extraction (DBE) tool in PixInsight.
First, using the tool, we select points on the master where there should be uniform background. Here’s my selections for this image. You’ll note that I tried to avoid areas where the dark gas cloud exists, but it’s pretty hard to judge on this image:
Then the map is created using the tool:
Then the map is subtracted from the master:
You can see the background is much more even now. It’s too light, but its much more even.
To clean it up, let’s apply curves again. Here I shaped using the Luminance and RGB levels and a small change to the saturation:
Obviously a lot more can be done to improve this image, but this is just an example to walk though the big steps. At this point I would probably adjust the curves a bit more, apply a little noise reduction filter to it, then save it in a 16-bit tif format and then do the final work it in Photoshop, The Gimp, or other photo editing program. At least we’ve got something to work with now.