FWHM and Eccentricity

[Eddy Timmermans]

I tried the SubFrameSelector tool on a series of images I took a while ago.
How do I use the information ?
Do I choose the frames with the lowest FWHM or eccentricity ?
Strangely enough, when I checked the frames with the highest and lowest FWHM values, they both had stretched star images in the middle of the frame.
Until now I mostly used the Blink tool, but with over a hundred images, it becomes quite tedious.
I found a number of topics here, but didn't find a clear answer.
So can anyone tell me how to select the best frames ?
THX
Eddy

[mschuster]

Hi Eddy,

IMO blinking is good to do even with a lot of frames.  Things like bad gradients, thin clouds causing halos around bright star, etc, can be seen easily by Blink but not so easily in the script.

For the script smaller FWHM and smaller eccentricity is better of course. I like to plot each and discard the larger valued outliers. How many to discard is a judgement call.

With stretched stars the script is still trying to do the best measurement it can. The ones with more stretch should have higher eccentricity values.

Thanks,
Mike

[jkmorse]

Eddy,

Here is an excerpt from my Cribsheet where I set out a number of ideas that others have posted here on the forum.  Its a great place to start though, Mike, if you take issue with any of these suggestions, that would be great to know since you are THE expert:

Step III – Image Selection using SubFrameSelector

Use the SubframeSelector Script to identify best original subframes.  Note the change in the process step order.  Per the SubFrameSelector documentation, evaluation and weighting is best done AFTER calibration and CosmeticCorrection but BEFORE StarAlignment. 

a.     One Option is: 
     i.     Approval:       FWHMSigma < 2 && SNRWeightSigma > -2
     ii.    Weighting:     FWHMSigma

b.   Another and probably better, formula is:
i.   Approval: FWHMSigma < 2 && SNRWeightSigma > -2 && WeightSigma > -2
ii.   Weighting: (2 * SNRWeight) – FWHM

c.   But the best, at least as to weighting, may be the one set out in the Trapped Photons tutorial I mentioned just below.  That formula is more labor intensive but uses actual data from the images themselves to develop a solid set of weighting criteria.

i.   Approval: FWHM < [4] && Eccentricity <  [0.665] && Noise < [120]
ii.   Weighting: ({35}*(1-(FWHM-[3.008]/[4.008]-[3.008]))+{5}*(1-(Eccentricity – [0.3663]/[0.5712]-[0.3663]))+{10}*(1-(Noise-[107.6])/[117.5]-[107.6])))*50

Those formulae, while looking intimidating, are actually extremely flexible.  They allow you to pick whatever parameters you want to focus on (feel free to change out any of the ones selected with your own choices).  Second, the numbers, at least under Weighting, are driven by your image data.  Under Approval, the bracketed numbers simply define how tight you want to go in managing your subframes.  In Weighting, the [xx] brackets define the high and low numbers for all of your images for that particular weighting factor.  The {yy} numbers add up to 50 and, when combined with the “*50” that closes out the formula, sets a floor of 50% on how much a weak but passing image is weighted.  And, again, you can change these {yy} settings around (so long as they still add up to 50).  I encourage you to read the Trapped Photons tutorial referenced below to get a full understanding of this powerful new way to look at your subframes.

By the way, if you like the sigma criteria from option (b) but the Weighting from option (c), feel free to mix the two.  It’s your data, do what you want.   

There is a lot of great information about the use of this tool in the documentation for the tool.  Also be sure to check out the following post, brought to my attention by Antoine who you can find regularly on the PI Forum:

http://pixinsight.com/forum/index.php?topic=6144.msg41751#msg41751

Note that you can often improve your master by eliminating weak subframes, particularly those that are blurry or suffer from light pollution effects and subframeselector, along with the Blink tool, will help you identify the flawed images.  It is painful to take out subframes you worked hard to capture but test both with and without the weak subs and you may see a significant improvement in the master.

Best,

Jim