Image Processing
Part I - Homework & Image Capture
This is the first in a series of practical explanations showing how I go about capturing and processing my images. There are about as many different ways of doing these things as there are astroimagers, and the guys who really know what they are doing have truly superior routines and skill sets. Still, these methods have worked well for me and may give you some food for thought, especially if you are just getting started (note that I have shamelessly borrowed techniques that work for me from those master imagers who have graciously shared their insights in magazine articles or online). Please drop me an email at jkmorse57@gmail.com if you have any follow-on questions (and never feel shy about asking even the most basic question, that’s how we all learn) or if you have had success with different ways of doing things. I am always ready to learn new and better methods.
-
Homework
- Prepare library of Dark and Bias calibration Masters
- Minimum 16 exps for each Master frame (I use 25)
- Bias Masters for each temp (I shoot 50 subframes)
- Two options for Dark Masters
- Long (30+ min) frames for each temp. (for Dark scaling)
- Frames matching light frame times for each temp. (I use both methods)
- Capture Option 1 (this option less viable for CMOS cameras since they don't actually bin your images):
- Binned 2x2 for RGB (assuming you bin your RGBs)
- Binned 1x1 for Lum, Ha, SII, and OIII
- Capture Option 2 (my method)
- Binned 1x1 RGB and/or Ha, OIII, and SII
- Under this method, I do not shoot lums at all, but instead create a synthetic lum as a part of my image processing from the RGB frames/
- Select Imaging Targets using CCD Navigator, AstroPlanner, your own favorite planning tools, or wherever you get your inspiration.
- Prepare library of Dark and Bias calibration Masters
Notes: Obviously, much of this applies to cooled CMOS cameras only. If you are imaging with a DSLR or a non-cooled CMOS, your only real choice is to take your Darks in the field so you match the temp of your lights. However you take your darks, it is absolutely critical that you match the temperature of your darks to the temperature of your light frames. If you don't, subtracting darks is worse than useless; it will seriously degrade your images. If you examine my “Learnings” page, you will see the results of skimping on getting the Darks right (and on not knowing how to do Flats, which is a necessary skill, especially for slower scopes (e.g., f/8) that we’ll discuss in more detail below).
This is initially a time consuming activity but what else do you have to do on cloudy or moon-filled nights? Besides, the Bias sets take no time at all (zero time exposures) and the Dark frames can be left on their own for hours on end (I set it up and let them run while I am sleeping). One thing to consider for cooled cameras is how to mimic indoors the colder temps you may be experiencing under the stars, expecially in the winter months. My solution used to put the camera in the frig but if you read my “Stupid Things to do While Imaging” page, it was while setting up a refrigerator run that I succeeded in dropping both my CCD and my computer on the floor. A better answer is to set up your system outside and shoot darks then or simply use a warmer temperature for your darks and lights. I always used to strive for -10C when I was shooting with a CCD, but CMOS have such low noise profiles that I now shoot at 0C. If you are using a quality CMOS your Dark and Bias library will be good for 6 months or more but be sure to check regularly and reshoot if you notice any thing strange in your processed images, such as banding. Any of the better processing programs makes creating Masters a breeze. I happen to use NINA for image capture and PixInsight to make the Masters, but really it’s just a matter of finding a workflow that suits you at a price you can afford (Nebulosity is a great piece of "starter" software in this regard).
As stated above, I use Option ii in creating my Master Darks. I understand the theory of Dark scaling and it works, but I like the extra control of matching my Dark exposure times to my actual light frame times in most instances. I also now only use Option 2 for my image capture. I have fully bought into the theory I discuss elsewhere that shooting 1x1 RGBs and then creating a synthetic luminance frame from them is the superior way to image.
A few thoughts on how I manage my files. Before I start imaging for the night, I set up my file folders. All of my image files are in the same master “Images” file on my system. The next level is that I have separate sub-files for each CMOS-OTA combination. I then have a separate file for every night of imaging. Within those dated files, I include a separate file for each target and a separate file for all of the flats I take that night. I keep all of my raw images for a particular target in that target’s file folder (e.g., all luminosity, red, green, and blue lights for a particular target are lumped in the same file (by “lights” I simply mean the images that capture a target and its associated star field, unlike the flats and darks that are for calibration only). Similarly, all my flats for that target go into one file subfolder within the relevant target folder.
If you decide to do the same, one convenient thing to consider is labeling each set of images of the same class by a unique code. For example, all of my red images of a target have the code “Rd” at the end, green has “Gr”, etc. (NINA and some other image capture software automate this process, making it even easier, and let you designate exactly what info is included for each light subframe). When I do my flat fields, I use a similar system, labeling red flats with a “Rff” code. (Note I did not put in “Rdff” since I want to be able to easily separate each series and any duplicate letters (such as the embedded “Rd” will confuse the red flats with the red lights, thereby defeating the purpose of this little exercise). Now, when I want to open all the red images in a series, instead of having to pick through the whole mess of files, I type *Rd* in the file identifier box and hit enter. The system now shows me just my red files. Then it’s a simple matter of holding down the shift + A keys to highlight the whole list and hitting open. Most people probably already know this stuff, but I find these kinds of shortcuts fascinating.
I used to recommend both CCD Navigator and AstroPlanner for planning imaging sessions and they are probably still excellent tools if you need them. Either one will do the job on their own but they offer different tools for getting to the same place and I find the combination of the two best suited to how my mind works (AstroPlanner is more database focused and CCD Navigator is more image driven). Now I am looking to shoot more obscure targets, such as dust clouds. The tool I turn to almost exclusively these days is the Sky Safari Pro 8 app on my IPad. It includes a very thorough catalogue of really obscure targets that is invaluable for me.
Another thing to keep in mind that applies both to your homework and your image capture is the type of file you save your images as. Whether you are using a DSLR or a CMOS camera, you will want to save your images in the FITS file format, initially in 16 bits when capturing and using 32 bit floating point for all processing. Then, if you use Pixinsight, you will want to adopt their native .xisf format. NINAS allows you to save you images to the .xisf format in the first instance. The key is to preserve every bit of data for the later processing steps. Once you start combining images, even images taken with a 12 bit DSLR, you can quickly exceed 16 bits of data. That is certainly true if you are using a 16 bit CMOS (of course its more complicated, everything is).
2. Image Capture (NOTE: these suggestions no longer reflect my way of doing things since I image remotely at Starfront Observatories. Still, you may be able to use some or all of these suggestions.)
- System Set-up
- Polar Alignment
- Focusing
- Autoguiding
- Learn to take Flats!!!
- New Flats every time you change the image train, rotate your camera or really, just breathe wrong.
- Expose until you are in the 18,000 to 24,000 count range
- Separate Flats for each filter
Notes: Your first task is getting the system set up properly for the night's imaging. The first critical undertaking is getting your scope polar aligned. Failure to get this right means your images will suffer field rotation around the edges. I discuss the problem in more detail and provide some examples in my Learnings page. You need to find a system that works for you. Drift alignment and other mechanical methods are full proof, if time consuming. I have gone the software route, using the Polar Alignment tool in various software suites. I get the scope aligned by eyeball when I get started. When I can start to see stars (but well before full darkness) its time to execute whatever protocol your software choice uses. There are simply too many options to address them here. Be sure you understand the software well, using a combination of the manual and YouTube videos or other online resources.
Focusing is the second critical setup function. You can do it manually, but even then be sure to use some helpful tools. The best is a program that will give you a full width half maximium (FWHM) reading which you seek to minimize. The best method, however, is to use an automated solution (NINA's HocusFocus is superb). As with polar alignment, you will need to spent some time getting a handle on how your preferred solution works.
The other setup item I recommend using is to dither your exposures. A program like NINA makes dithering a breeze and you will appreciate having done so when you see how adding dithering helps eliminate hot pixels when you are stacking your images. And, if your images are undersampled, dithering allows you to access drizzle routines that are almost magical in how they improve an undersampled image.
Though I started out trying to capture up to 6 targets a night (a typical beginner’s mistake), you simply cannot do justice to more than one or two images a night. I currently shoot one image over many nights but that is the luxury of a remote setup like Starfront. The gains in SNR are more than worth it. And besides, I am not giving up the hobby any time soon so what's the rush.
Let’s spend some quality time talking about Flats (now would be a great time to slip over to the frig for a favorite libation; don’t worry, I’ll still be here when you get back). Again, on my “Learnings” page I show what happens if you do not use them. Flats cure two important flaws, vignetting at the edges of the image (actually a rather minor annoyance that can be remedied in other ways) and dust donuts marring the frame (a major headache that can ruin a set of otherwise excellent images). You can live without them, though using that logic, you can live without Dark frames as well. But if you are serious about getting everything you can out of your images, you really need to master this skill (and really, it’s not all that painful).
There are a lot of articles that deal with Flats but they did more to confuse me than to explain how to take Flats competently. More than likely my fault, but they often got lost in the science and not the process. They also tend to focus on the mysteries of Sky Flats. I am certain there are those who have mastered the technique and love Sky Flats, but they really require a level of art that is beyond me. I use the mechanical solution instead. I started by building my own contraption, using opaque “milk carton” plastic and a shielded hanging work lamp. Now there are lots of mechanized flat panels that fit right on the end of your scope. I use a product from WandererAstro, but there are lots of options.
Another option is to learn to shoot sky flats. They require no extra equipment, except maybe a white piece of cloth. Again, it is a skill you will need to practice to become proficient but there are lots a resources to help you get started.