Image Pre-Processing

So let’s get started.   First a word on nomenclature.  When you are out in the field, you are engaged in image capture.  These next steps describe “pre-processing” meaning getting the raw sub-exposures cleaned up and combined into image stacks.  Depending on how you look at it and what package you use, pre-processing may end and processing may start once you have your L, R, G, & B stack processed into “final” L, R, G, and B combined images or after you have created your RGB image and have a “final” RGB image and a “final” Lum, ready for combination and further processing.  However, in PixInsight, for example, where you can do everything in the same package, there really is no clean break point.  There the “breakpoint” is when you convert your components from linear images to non-linear ones through stretching.

In this first section we will deal with pre-processing an LRGB image (LRGB stands for luminosity or Lum, red, green, and blue image components, respectively).  Note that if you are using a DLSR or other “color” camera for your imaging, the calibration steps will be the same, but there are no separate LRGB images to combine later; unless of course you break them out and create separate L and RGB channels to play with.  While I used to do that with my Canon 40D images, I will not be covering that here.  For HaRGB and narrowband images, assume that the same processes apply except that you substitute the Ha image for the Lum for HaRGB images and the SII, Ha, and OIII images for the RGB images for narrowband images using the Hubble palette.  In the latter case there usually is no Lum image as such.  Also, if you are interested in following my lead and eliminate shooting separate Lums, you will not have Lum sub-exposures, though if you follow my PixInsight routines, you will be creating a synthetic Lum from the R, G, and B stacks.  Another thing to remember is to save your work regularly.  You can delete all of the interim steps to save disk space once you are done, but you will be glad you don’t have to go back and repeat all of your steps if you make a mistake somewhere along the line and need to do that step over.  Some software packages let you save the history of your workflow so you can often automate this step.

If you have done what I recommended in Part I, you have a series of red, green, and blue (and possibly Lum) images in the file folder for the target in question.  If you are following my routine, you also should have at a minimum, a series of darks, taken at the same temperate, the same binning mode and the same exposure time as your lights (if you took all your RGB images at the same temp, the same binning mode, and for the same exposure time, you only need one set of darks, not three).  You may also have a set of flats for each different image set (unlike the darks, here you will need different flats for each color even if everything else remained the same).  You may also have a set of bias frames (in PixInsight I actually get better results calibrating my flats using bias masters but others prefer to use master darks that match the flats).  Note I do not intend to cover scaled darks here, but drop me an email if you need more information).  

We now need to take that lump of image files and convert them into a stunning astro-image.  The first step is to build your master darks, and if you took them, master flats.  The steps are the same for both expect for one added element for flats.  Most astro-image processing programs make this a semi-automated process (or in the case of PixInsight and MaximDL, you can fully automate it though I find that I get better results if I take the time to do each pre-processing step manually).  The goal is to combine all of the dark files for one set of images into a master dark.  The typical setting for the combination is “Average”.  Do the same for each set of darks.  For flats you do the same, but can also subtract “flatdarks” or the master Bias for the flat frames.  You make the flatdarks at the same time and using the same settings you used when creating the flats.  In PixInsight I get my best results by only subtracting a master bias but you should try both methods to get the cleanest flat master possible.  If you are going to follow me into the wonderful world of PixInsight, be sure to use my cribsheet as a reference for your preprocessing until you get comfortable with all the permutations.

Once you have the masters in hand, next you apply the masters to your image frames.  Open all of the red image frames, go to the calibration tab and select the appropriate flat, dark (and if you have one, bias) master frame.  Hit the button and the program does its thing and you now have a set of fully calibrated images for the next step, registration and stacking.

Again, different programs do this different ways and in different sequences, but the main components include aligning (or “registering”) the images, equalizing them, doing a data reject, and them combining the images into a “final” product that is ready for further processing.   Read your manual or help files to see exactly how you do each of these steps for your program of choice (if you are using CCDStack 2.0, take a serious look at Adam Block’s tutorial video and if you are using PixInsight, look at Harry’s tutorials or check my cribsheet).  You will need to repeat these steps for your red, green, blue, and luminosity file sets.  At the end, you should have a set of 4 “master” LRGB images ready for the next step.

Now its time to create your color image.  First, open the RGB images.  If you cycle through them, you will see that they are not aligned.  This will be especially true if you followed my earlier advice and dithered your image capture routine.  As such, you will now need to use the same routine you used to align or register the individual color files.  And since you will be combining the RGB image with the lum in a later step, you should go ahead and align it at the same time (aligning the Lum is not necessary if you are creating a synthetic Lum in PixInsight since it will be built from the aligned R, G, and B images).   Note, however, if you shoot both Lums and RGBs and binned them differently, (e.g., RGBs binned 2x2), you will need to resize the RGB image to the same size as the Lum image before they can be combined.  Just double the size of the RGB images and that will take care of matters.  Now open the “color” tab in your program and assign the RGB images to their respective red, green, and blue channels.  Hit the button and . . . unless you have normalized the color stacks, you probably have an image that is some bizarrely off color shade.  Not to worry.  You are usually afforded the opportunity to reset the background color which should bring things a lot closer in line with expectations.  If you are really ready to go to the next level, you can read up on color correction using G2V stars to get the necessary adjustments to achieve a true “white” setting for your filter set (though, again, PixInsight offers a cleaner and easier way to achieve the same result).

Now that you are done with pre-processing, its time to enhance your image in the processing step.  There are several ways to tackle that adventure.  As I have indicated many times now, I used to do my processing in Photoshop after dong my pre-processing in CCDStack 2.0, but now do all of that work in PixInsight.  If you considering using Photoshop and have not yet made the investment, I understand GIMP does many of the same things and is available for free download.  If you are stopping here, you will need to add your luminosity image as a part of creating your color image.  But the flexibility you gain by processing your images is so great that you will want to explore going that route eventually.  This is particularly true if you are attempting HaRGB or narrowband imaging since you will want the extra power provides to manipulate those more complex images. 

If you are going to try PixInsight, take a look at my cribsheet and Harry’s tutorials.  If you are going to tackle Photoshop, then I strongly recommend the Adam Block tutorials.  Either way, enjoy the journey and let me know how things are going.  I would love to hear how you are progressing. 

 

Next Up Part III - Image Processing