DSS and PI FWHM values?

[ChrisEll]

My guiding has steadily improved over the past few weeks but my DSS FWHM values tended to suggest only modest gains, if any. I recently used PI to examine the lights and was surprised to see they were so different...?

 

[tooth_dr]

This is interesting to see. I’ll be following this thread 

[Adreneline]

2 hours ago, ChrisEll said:

FWHM values

I am sorry I cannot shed light on this but I also will be interested to see how this thread develops. FWHM values mystify me and I too have noticed differences in the way different software produce differing values from the same images. And not wishing to de-rail this thread but why is PinPoint required by SGPro in order to use FWHM as an alternative to HFR (which SGPro claim in a more robust system). I've got to the point where to minimise stress (and maximize imaging time) I ignore the absolute values and just look at trends and aim for a minimum - whatever it might be!

Thank you for starting this thread.

[davew]

It is my understanding, from what I've read in the past, that DSS analyses the whole image and most of the others don't.

That means DSS will look at those flying bats or bent sausages in the corners of your image and try to make some sort of sense of them

Dave.

[tooth_dr]

4 minutes ago, davew said:

....those flying bats or bent sausages in the corners of your image

Dave.

Made me LOL.  I'm using that excuse for a poor FWHM

[sharkmelley]

I would take a couple of very different subs (e.g. DSC_0533 and DSC_0540) and run the FWHMEccentricity script on them, producing plots.  See if you can correlate what you see back to the FWHM in the SubframeSelector.  By the way, what model does the SubframeSelector use:  Gaussian, Moffat, Lorentzian?  Or does it use the one that appears to give best fit?  Presumably DSS uses Gaussian?

Mark

[ChrisEll]

FWHMEccentricity gives practically the same values as SubframeSelector. PI uses Moffat4. No idea about DSS. 

[Stub Mandrel]

What do they measure the FWHM in?

DSS knows nothing of focal length so it calculate in pixels, I assume.

If PI knows the system setup it may calculate it in arcseconds, which will often be the same order as the pixel value, but not the same.

[sharkmelley]

On ‎13‎/‎03‎/‎2018 at 13:07, ChrisEll said:

FWHMEccentricity gives practically the same values as SubframeSelector. PI uses Moffat4. No idea about DSS. 

I expect DSS is using Gaussian.  So set PI to use Gaussian and the values might then be similar.

Mark

[ChrisEll]

I ran the test with Gaussian and Lorentzian methods. The numbers for Gaussian are nearer the DSS values but still a long way off. There was some difference in e as well.

[sharkmelley]

Your stars have a medium to high eccentricity.  So the FWHM in one direction is higher than the FWHM in the other.  So which FWHM should be reported?  The minimum, the average or the maximum?  Notice how the larger the eccentricity, the larger the discrepancy between the DSS FWHM and the PixInsight FWHM.  It's simply a question of which FWHM to report when the stars are not round and DSS does it differently to DSS.  That's all. There's no right or wrong.

 

[Later Edit]  I should have spotted that earlier!

Mark

[ChrisEll]

Ha, yes, I've yet to add a flattener to my OT. My stars do get quite eggy at the edges.