Astro Stacking

"Stack" images to synthesize longer exposures on the D7/ug/i

Introduction: Since the D7 is limited to 30 second exposures, it is necessary to find some way of synthesizing longer exposures for astronomical work. A lot has already been written about "stacking" astro images, and I certainly can't repeat that all here. The technique that I describe here is slightly different than what has been done in the past (as far as I know) in that it allows each image to contribute to the final image equally, and allows the noise to be averaged out to the maximum possible.

The sample of M57 above shows the result of using this technique on 20 different identical 30 second ISO 100 images. The only tone correction needed was a very slight reduction in the very lower end of the tone curve to remove sky glow.

How the stacking is done:

 To stack these exposures, I used PhotoPaint 9. You can use any photo program that allows you to work with layers and perform math operations though.

The overview of the procedure is this:

* The idea is to divide the available exposures up into equal groups. Each individual group contains as many exposures as you can add together without having values for your object of interest go over 255 and "clipping". In the sample case above, I determined that about 10 exposures could be added together without going over 255.  I did this by passing the cursor over the nebula in a single exposure and noting that the highest value was 25 (25x10 = 250). This gave me two stacks of 10 images. You might have to discard some extra images if your number of images does not divide equally into the number of stacks.

* I made two combined stacks of 10 exposures each, then reduced the range in each stack to run from 0-128 and added the two stacks together. The end result is a final stack where each exposure has contributed equally, with the values in that stack running up to 255 but not over.

* I stacked exposures by starting with one exposure, then cutting and pasting subsequent exposures as new layers. Each layer was aligned manually at 200% view.

* I used tricks to help align the layers. This is necessary because it is difficult to tell when you have perfect alignment of a faint new layer with a bright stack of images.  For the first two layers, I set the layer operation temporarily to XOR, then aligned for minimum brightness on stars. Switch the layers back to "Add" then combine once aligned.

* For subsequent layers I aligned with the "exclude" operation. This gives a small dark spot generated by new layers in the center of bright stars that you can position right in the center of a star image. Again, don't forget to switch the layers back to "add" before combining.

* Keep adding until the values on your subject of interest approach 255 (if you calculated correctly you will be OK). In no case add in more than half of your images if adding all will result in values over 255.

* Repeat for the other stack of images.

* Now transform the images (assuming 2 stacks) from 0 to 255 to 0 to 128 range.  If you had three stacks you would use a range of 255/3 or 0 to 85.

* Pick a stack and cut out the other stacks and paste them on after aligning. For the first two stacks since they will be equal brightness you can use the XOR operation trick above to align, then switching to add.

This results in a perfect image with perfect range, and with all images contributing equally and with noise averaged as much as possible. The sample shown above had only very slight adjustment of the lower part of the tone curve only to remove sky glow.