Which subs to keep?

[tooth_dr]

This has been niggling at back of my mind for some time now, so I thought I had better ask.

 

How do you decide which subs to keep and which to not include in the stacks.  I'm not talking about obvious bad ones, but subtleties.  For example I have 94 x 300s lum subs of M42 currently being stacked in APP.  When I order them in terms of quality after they have been analysed, to my eye, sometimes the highest or lowest scorers don't look as good or bad as others in between?  But the software must be ranking then based on something?

 

Is there a way to establish the best % to stack?  Or just stack them all??

 

[Craney]

Have you run the same set of subs through DeepSkyStacker to see if it agrees with the ranking order  ??

I imagine it is all to do with the algorithm priorities...  and they are probably hidden deep in the coding.

94x300s  !!!!.... blimey,  the actual Nebula will have changed shape in that time.     Maybe that's the reason....... 

[Demonperformer]

Sharpcap live-stacking allows you to determine the basis on which they will be accepted/rejected into the stack. I just use FWHM.

[tooth_dr]

6 hours ago, Craney said:

94x300s  !!!!.... blimey,  the actual Nebula will have changed shape in that time.     Maybe that's the reason.......

That’s just luminance. Then there’s RGB and the shorter core exposures. ?

[almcl]

No knowledge of APP and not a lot about DSS, but three factors which appear to be involved in ranking the 'best' individual subs are star roundness, perhaps as measued by FWHM, sky background and number of stars. 

Sorting subs by each category in turn and eliminating the worst ones is a method I've used, but have also noticed that occasionally really poor ones slip through.  Not a huge problem when sorting 30 or so subs but maybe more time consuming if you're dealing with 100s? 

[x6gas]

It's a good question and I'm interested to hear what others do.

I generally stack in AstroArt and use the option to visually confirm each sub (you can ask the software to reject subs based on different criteria) but sometimes if I know the data is reasonably good I'll just stack the lot and let the stacking algorithm sort it out.  In DSS I sort by score and make a determination of where to draw the line based on what the scores look like and the integration time I'm aiming for.

I'm just experimenting with the trial version of PixInsight and trying to determine if I can see a difference in the stack if I've sorted by FWHM or eccentricity, for example.

My guess is that with a large number of subs it doesn't make much difference given the averaging that takes place during stacking but of course that not be the case if you have a good number of subs exhibiting a similar tracking problem...

[Knight of Clear Skies]

I mostly look at FWHM and reject outliers. Stacking with a sigma clip seems to be good for rejecting rubbish, satellites, planes and even a bit of light cloud doesn't appear to damage the final image, as far as I can tell.

[RolandKol]

Pix has nice tools for that...

1) Blink - to reject the obvious ones, with bushes in the FOV or very shiny U F Os.

2) Not decided which to use as the second yet, - SGP has a free Sub Grading tool, very easy to use, load press start, go for coffee... Or Pix SubFrame selector, which looks like a bit more accurate as takes into account not only FHWM but also other factors which you are able to tweak and include into a final quality score and which also can be written into each sub-frame data and used by Pix for later stack... Ofcourse, - Pix tool is MUCH more difficult to use, but it looks like it produces much better results... 

[tooth_dr]

Thanks a lot guys for the replies.  APP has a quality tab, as well as lots of others.  I must look at this closely later.  I was talking about the 'QUALITY SCORE' heading in my opening post.

 

[RolandKol]

try free SGP app from here http://mainsequencesoftware.com/Releases/DownloadFile?id=4

It grades in Bulk, based on FHWM and amount of stars found