Hey folks,
I have been itching to get my head around the SHO script for some time and this post:
http://pixinsight.com/forum/index.php?topic=8706.0 finally pushed me into action. I wanted to see a comparison of using traditional NB linear combination methods vs using SHO on non-linear Ha, OIII and SII images. I had just taken a new data set of NGC7000 recently that seemed the perfect test case. Please ignore the quality of the image, it's too noisy (accidentally shot it at +10C rather than -10C
) and was using a guider with a 15 arcsec/pixel resolution vs my imaging CCD's 2.59 arcsec/pixel so the stars are out of round
). Still the underlying raw data is excellent, made up of 36 subframes in each of Ha, OIII and SII (180sec iterations on my f/3.3 Takahashi e130D).
I started using my traditional method of cleaning and mixing the NB image in the linear phase, using a SynLum made up of 100% of all three, and doing the channel combination in PixelMath using 80% SII and 20% Ha for Red, 80% OIII and 20% Ha for Green, and 100% OIII for Blue. Here is the end result of those efforts:
http://astrob.in/full/198688/0/Not bad, and had I known nothing about SHO I would have been pretty happy with the result. In fact, I would have been thrilled had I not seen what SHO could do.
To set up the SHO test, I did everything I normally do for the three NB stacks, except I did not combine them in the linear phase, instead doing an HT stretch on each of them, trying to maintain the same flux (should have thought to do a Linear Fit after the stretch but didn't think of that at the time. Besides, this is only a test. I then populated the SHO script using the same splits I used in the traditional method, namely 80% SII and 20% Ha for Red, 80% OIII and 20% Ha for Green, and 100% OIII for Blue. I also ran it three different ways from there. First, following the advice from the post referenced above, I built a Lum image using just Ha at 100% and SII at 60% (using the lighten option). That produced the following:
http://astrob.in/full/198690/0/Ok, I don't know about you, but that image simply blows me away. I think its incredible, notwithstanding the flaws I noted above. And that was all done by the script. All I had to do was input the data and push the button. Simply amazing results. What a script!!
In the second test case, the only thing I changed in the SHO settings was to build the Lum using 100% of all three stacks. Here is the result:
http://astrob.in/full/198691/0/I can find no difference from the image using just Ha and SII to build the Lum. If you see a difference that I can't catch I would love to hear about it. Finally, I adjusted the SHO settings to do the combination without creating a Lum image. Here is the result:
http://astrob.in/full/198693/0/No question, that is clearly a step in the wrong direction. The script makes superb use of the SymLum it builds so I strongly recommend always going through that extra step.
Finally, I wanted to test it on a second set of data, just to see the result. Again, this is not perfect data, shot with a Canon lens and manual, therefore less than adequate focus. And the results are not as clear cut. The first is the traditional image I have posted in my public gallery on Astrobin:
http://www.astrobin.com/197380/Its a nice image given the quality (or lack thereof) of the subs. And here is the SHO result:
http://astrob.in/full/198687/0/I think that one is better, showing a bit more detail in the nebulosity, but I couldn't get the stars fixed and that distracts from the image since its such a crowded star field. But not a waste either. One thing that could be making all the difference is that I shot NGC7000 during a new moon and IC1396 during a full moon. Even NB imaging isn't safe from that orb.
In any event, I hope you find this exercise useful. I certainly did, and from this point forward I will be running my NB images through SHO as my starting point. Next I plan to test whether SHO can do some its same magic with RGB images. I have heard that its best to stay away from SHO for those, but if I find out differently, I will be sure to let you know.
Best,
Jim
