FWHM and Binning

[Rodd]

This is not about FWHM and binning as they relate to one another...rather, this is about FWHM, and also about binning in software as a separate, unrelated issue.  

FWHM

I have always fretted over FWHM,  Throwing out subs if the FWHM is to high (over 3), wondering if my guider resolution was the cuase of elevated FWHM, and on and on. Its results like this that tend to confuese me.  The FWHM of the unbinned Ha stack is 3.3--up until now I have considerd this unworthy of imaging.  But it does not look bad to me at all with respect to FWHM.  So, my question is...how important is FWHM.  What should be regarded as to high for use.....obviously 3.3 or even 3.5 (maybe 4?) is not too high.  If it was, I would have thrown out all this data and from the looks of it I think its a pretty decent start of a Thor's Helmet image.  Then again--images by themselves often look good even if they are not nearly as good as images the viewr has previously seen--I think thats a byproduct of our eyes and brain.  Side by side with a stack 2.0 FWHM this image  may, in fact, look pretty poor indeed.  Hard to justify throwing this start out, however.  My concern with FWHM is not star size--its the fact that fine scale detail may not be visible if FWHM values are higher than acertain point....what that point is, is the purpose of this post

 

Software Binning 2x2  

I now see the benefit of software binning.  Previously I was only looking at the screen stretch versions of binned and unbinned stacks, and they look identical.  That is becuase the screen stretch function calculates a stretch and utilizes a different stretch for the binned and unbinned stacks.  So the look teh same.  However, when you render the stacks non linear--even with an identical histogram stretch (manual, not using the STF or other automated tool), the difference can clearly be seen.  You be the judge.  I suppose i am posting this for those folks on teh fence about software binning, or not aware of it at all.  It is teh same as hardware binning with CCD (or very close).  But many CMOS cameras do not support hardware binning, so software binning is the only option

This stack was talen with the TOA 130 and ASI 1600 with a 3nm Ha filter: 14 300 sec subs  (I through out 10 wwith elevated FWHM values....should I have?).  Pardon the quality of the background--there are only 14 subs.  Quite frankly, I am surprised the background is not a lot worse considering this was taken during a 96% Moon.  This was an add on project while I waited for M101 to climb above the trees--so the camera angle is verticle as opposed to horizontal.  Its hard to know if resolution was lost becuase the binned stack is so much brighter.  Details too dim to show in the unbinned stack are showing up in the binned stak by virtue of thie brightness--not necessarilly due to their higher resolution.  To my eye, I can see an indication of a slight loss of detail--more than made up for by teh signal strength.  The images have not been processed much--just a wee bit of noise control to push the histgram away from the left hand edge so I could drop the black point appropriately (very low NC settings).  Other than that, just crop, DBE and a stretch--being very careful to make sure the stretches were the same.  Not that easy, becuase the shapes of the histograms are slightly different- But they are as close as is possible to get them with the left hand edge at precisiely the same spot.  More data is obviously needed to improve the background.  Anyway--not sure I have the stretch right--after 6 attempts I figured I'd go with what I have.  Personally--I like the unbinned

Unbinned 14 sub stack

 

 

Binned 2x2 in software

[tooth_dr]

Depending on what stacking software I use, I get different values for the FWHM, from the same data.  

[Rodd]

1 hour ago, tooth_dr said:

Depending on what stacking software I use, I get different values for the FWHM, from the same data.  

That’s interesting.   How much variability?   Very different?