FWHM and Brightness filter for frame selection problem

Anything that doesn't fit into any of the other forums
Post Reply
timh
Posts: 690
Joined: Mon Aug 26, 2019 5:50 pm

FWHM and Brightness filter for frame selection problem

#1

Post by timh »

Hi Robin

It is very useful to be able to select which frames to save and which to reject during capture. It enables leaving capture running even during nights of broken skies with clouds moving over while saving on data storage and downstream processing. I have been in the habit of using the FWHM and Brightness filters.

However the two measurements are clearly not orthogonal and the measurement of FWHM moves up and down with brightness in a consistent way -- in fact I'd estimate that - in a quantitative sense - most (60-80%?) of the variation in FWHM mirrors that in brightness. So by putting in a cut off value for FWHM I suspect that the unintended effect has been to exclude potentially the best and brightest frames?

The next stage on from Sharpcap selection is (for me anyway) to use Subframe selector in PixInsight to select the best frames on the basis of the FWHM and - a function that the PI developers have been gradually evolving and improving - it combines he most critical quality factors (signal, noise, PSF dimensions and shape, background gradients) in a single function that sort of expresses how well correctly shaped sharp PSFs emerge from the murk. Not surprisingly -- in this case high values of PSFSignalWeight tend to correlate with low values of FWHM -- so by itself it is usually probably all the selection that is needed cf FYI

https://pixinsight.com/doc/docs/ImageWe ... 20function.

For Sharpcap wondered whether it might make sense to try a selection parameter that was some combination of Brightness and FWHM ? Brightness already works well by itself as a selection criterion --- maybe FWHM/ brightness would provide a sort of 'brightness-normalised' FWHM value ?

Tim
Attachments
BrightnessvFWHM.jpg
BrightnessvFWHM.jpg (75.58 KiB) Viewed 852 times
User avatar
admin
Site Admin
Posts: 17584
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: FWHM and Brightness filter for frame selection problem

#2

Post by admin »

Hi Tim,

I have my suspicions about what is going on here - currently the set of stars that are included into the brightness and FWHM measurements in each frame just depend on what stars get picked up by the star detection algorithm in that frame. Stars that are too saturated are ignored for FWHM, but other than that, the figures can be affected by changes in detected stars from frame to frame. Random noise will have a bit of an effect, but changing conditions will not only affect the things we are trying to measure (brightness and FWHM) of individual stars, but also the selection of stars that are included in the measurement, and I think *that* is the effect you are seeing.

The plan I have to deal with this is to stop the individual measurements from being completely independent of each other - instead track the stars detected in each frame and how individual stars change in value for FWHM/brightness from frame to frame. In this way, a correction can be applied to the mean values generated to try to remove the effect of different stars being selected. This is similar to how the current implementation of the 'Stellar Photometry (Experimental)' works. I can't just plug the new algorithms straight into live stacking though - the current algorithm is limited in some respects but works 'well enough' for 1000s of people. I don't want to introduce a new algorithm that is better in theory but somehow breaks things completely for a significant number of people, so I need to be pretty cautious about how this gets implemented.

Still, with the planets moving out of view soon, I will be spending less time on the planetary stacking/focus/things like that, and should have a bit more time to focus on some of this deep sky stuff.

cheers,

Robin
ChrisR Oz
Posts: 224
Joined: Tue Sep 28, 2021 11:03 am
Location: Melbourne

Re: FWHM and Brightness filter for frame selection problem

#3

Post by ChrisR Oz »

Hi Robin,

I see the same correlation, where as a star dims it’s ‘apparent’ FWHM decreases. Seems to be a systematic error in the FWHM fitting. This is why I use single star mode for focus scans, because the multi-star (as it drops and picks up stars as a function of focus) introduces random scatter to the curve (at least for my focal length/star selection).

There are similar issues in X-ray spectra fitting. Years ago I adopted a modified weight for least squares fitting that uses 1/f (f is the fitting function) instead of the usual 1/sigma-squared. The latter can be shown to introduce a systematic error to peak areas around chi-squared. I never investigated the effects on peak width. But this may help.

Cheers, Chris.
Celestron EdgeHD 8, Focuser; iOptron CEM120 mount on pier; Starlight Xpress AO, OAG and Filter Wheel; ZWO 2400MC, 294MC/294MM Pro and 174MM mini; SharpCap Pro, PHD2, Televue Powermate 2x, Baader Neodymium, Astronomik CLS-CCD, ZWO UV/IR, Duo filters
timh
Posts: 690
Joined: Mon Aug 26, 2019 5:50 pm

Re: FWHM and Brightness filter for frame selection problem

#4

Post by timh »

Thanks Robin,

Yes that does seems likely - and indeed now that Chris mentions it -- the stars that are selected can also change while using the multi-star FWHM -based focusing tool. So anything you can do eventually would be great. But in the meantime --- especially given awareness of the problem and modifying slightly the way they are used (i.e selecting frames primarily based on the brightness filter with the cut off for the FWHM filter set high to catch just the real outliers) the existing filters do a good job.

Tim
User avatar
admin
Site Admin
Posts: 17584
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: FWHM and Brightness filter for frame selection problem

#5

Post by admin »

Hi,

thanks for the suggestions and feedback (Chris and Tim) - oddly this brightness sensitivity mirrors a similar issue with the planetary focus measurements where the measured contrast is affected by image brightness and it turns out to be *very hard* to cancel out that sensitivity in a useful way (I've tried several approaches and only managed to reduce rather than eliminate it).

I can fairly easily run some tests to see how the current FWHM measurement (which is actually closer to measuring half-flux-radius) responds to variations in brightness of a sample frame - aside from the difference in *which* stars are detected as the brightness changes, I would be hopeful that the measurements for stars that are consistently detected should be fairly constant regardless of brightness.

Out of interest Chris, what sort of focal length/pixel scaling/etc are you using that suffers with issues in the multi-star FWHM? I find that one is robust for my usage.

cheers,

Robin
ChrisR Oz
Posts: 224
Joined: Tue Sep 28, 2021 11:03 am
Location: Melbourne

Re: FWHM and Brightness filter for frame selection problem

#6

Post by ChrisR Oz »

Hi Robin,

That’s at 0.64 arc-sec per pixel image scale and a FL of about 1500 mm. Mostly not very busy frames, as I’m often shooting out of the Milky Way at more distant curiosities.

Cheers, Chris.
Celestron EdgeHD 8, Focuser; iOptron CEM120 mount on pier; Starlight Xpress AO, OAG and Filter Wheel; ZWO 2400MC, 294MC/294MM Pro and 174MM mini; SharpCap Pro, PHD2, Televue Powermate 2x, Baader Neodymium, Astronomik CLS-CCD, ZWO UV/IR, Duo filters
User avatar
admin
Site Admin
Posts: 17584
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: FWHM and Brightness filter for frame selection problem

#7

Post by admin »

Hi Chris,

thanks for that - not far off the sort of scales I get with my 1800mm F/L classical cassegrain and 26C camera, so I will give that a try out when we get some clear sky. It usually works out OK for a focus scan, but normally I'm not really watching the fine details, so maybe I am missing something.

cheers,

Robin
Post Reply