Astrophotography - Processing
The goal of this page is to show my workflow in image processing. From this site you already know what equipment I have and how I use it, you can also browse my images in my image galleries. Image processing is the link between images and equipment - and hopefully it may help somebody to devise better and more efficient workflows .
The first step of processing is data reduction. I collect subexposures using two laptops, but all the images are transferred to my desktop for processing. The data reduction step is done in Nebulosity. Data reduction means removing artificial signal coming from hot pixels and dark current (using bias and dark frames) and correcting uneven illumination with flat frames.
There is a lot of information available about data reduction, image pre- and post-processing, both online and in books, so I will not repeat it here. Nebulosity help contains lots of examples and tutorials that explain what is done and why, there is also extensive help available on Stark Labs web site. As for the final processing the source I like best is the book "The New Astro Zone System for Astro Imaging" by Ron Wodaski and Russell Croman. I strongly recommend reading both (i.e. Nebulosity help and the book), it will save you many unnecessary frustrations! Another great source of post processing information are IP4AP tutorials.
In order to carry out preprocessing (data reduction) one needs master bias, dark and flat frames. The best is to produce them earlier and set up a library of these files. I prepared master dark frames by averaging 40-60 individual frames with strong sigma clipping; master bias frame was created from 100 subframes. Some cameras are better suited for dark frame using, some are better off with hot pixel maps. For more info see cameras page. Flat frames need to be prepared for each optical configuration, so it is necessary to prepare them immediately after imaging session when new optical configuration was used - I do it with the help of a light box. I don't shoot flats after each session, only when optical configuration changes (new reducer, different OAG orientation regarding the sensor etc). All my darks and flats are bias-corrected.
First step in preprocessing is applying darks (or hot pixel maps), bias and flats. Then images are normalized and debayered into the color fits. Color images are then aligned (with translation and rotation), but with the option "save each file" switched on. Pre-aligned files can be then stacked with sigma clipping (removing outliers), usually with 1.5 sigma threshold. The stacked image is saved as fits and exported to 16 bit TIFF.
The next steps of processing take place mostly in Photoshop. First, standard stretching curves are applied (see the "NewAstro Zone" book!), with levels and slight color saturation boost (15% - to preserve colors that are diminished by curves). After first few curves colors are balanced using RGB levels tool. Gradually less aggressive curves are applied followed levels. At this point depending on the image it may be needed to remove gradient with GradientXterminator and eventually it may be useful to smooth out noise with Neat Image (or Gaussian blur on dark areas). Finally contrast is enhanced in bright areas with high pass filter method or unsharp mask . In the end it may be necessary to further boost color saturation and adjust color balance.
High pass filter can be done in Photoshop by duplicating current layer, and setting the new layer's overlay method from "Normal" to "Soft Light". Then High Pass Filter need to be applied to the new layer (Filter->Other->High Pass), usually the radius around 10 pixels works fine. When the result is satisfying the image can be flattened. Unsharp mask is more traditional way of sharpening, one need to be careful to choose appropriate pixel threshold to avoid sharpening the noise, or even better apply sharpening to bright areas only.
There is a set of Photoshop actions that perform many standard tasks (like star selection etc) available for purchase from Noel Carboni. I use some of them from time to time, especially star selection. It is often a very good idea to separate stars into different layer to avoid over sharpening them or just to apply certain modifications to them. These actions are essentially shortcuts, an advanced imager may do the same thing manually, if (s)he knows how .
In the near future I will add tricks and techniques for various aspects of postprocessing I often use as sub-pages (menu on the left).