Puzzled about FWHM values when changing from Bin1 to Bin 2

[tim22]

Hi folks,

This is my first post - apologies that it is such a dry and technical one 🙂 -   and I was wondering if anyone could advise me where I am going wrong with the following  or explain what seemed  puzzling to me  relating to the effect of binning on FWHM values?

Using an SW Skyliner 250 mm F4.7 (f 1200mm) goto Dobsonian with a Baader coma corrector  and an ASI294 PRO MC camera (4.63uM pixels).   

I tried an experiment with 3s subs (gain 200, offset 29) to photograph star clusters (M34 and M103) in order to compare the results unbinned and at 2x2 binning,   At 3s there is relatively little added blurring due to Dob field movement.  The 2X binned and 1X images were taken within minutes of each other (so no substantial change in seeing or drift in focus).

The image scale for my setup is (I think)  ~ 0.8 arcsec   per pixel  unbinned  and 1.6  when 2 x binned.    Based on DSS scores  the FWHM -pixel   of 1X raw FIT files averaged about 4.2 pixel.   Thus I calculated that the FWHM in arcsec  was about 3.4  (~ SQR of sum of squared contributions from optics, movement and the seeing). 

What puzzled me was that when I went to 2x2 binning DSS reported average FWHM- pixel scores of about 3.2 pixel where as I was expecting perhaps about 2.1 ?  i.e.  with 2 x 2 binning the overall FWHM in arcsec would now appear to be about 5.1 and yet, relative to 1 x1 binning nothing in the optics,  field movement rate  or actual seeing should have changed?

There is obviously something that I am failing to understand here and I would be very grateful for any enlightment that folk might be able to offer

(the images themselves were not bad with the 1x images looking slightly sharper but less bright than 2x binned)

best wishes

Tim H

 

 

 

 

   

[Martin Meredith]

Hi Tim

Welcome to SGL! You might get better responses if you post in the imaging section but I'll have a go at a possible answer. Is it possible that DSS knows your that images were binned (this info may well be in the FITs header,  depending on which capture software you used). If so, the FWHM estimate might already have taken binning into account.

I guess I would expect FWHM values to be independent of binning so long as your arcsec/pixel resolution is well below FWHM in arcsec.

Martin

[vlaiv]

I think that DSS is not that good in determining FWHM values of the stars.

What happens here is that Gaussian shape fitting depends on estimated noise floor. Binned image will have lower noise floor.

Perhaps try using AstroImageJ to measure FWHM? I found that it is much more consistent.

Another important point is. What are your subs like since we are talking about OSC camera? Is bin x2 in drivers producing monochormatic image? It is generally advisable not to bin CMOS subs in drivers, but rather use software to bin your subs later.

OSC camera needs to be debayered first, and regular interpolation debayering + binning afterwards will not produce proper binning results - it will lower resolution but SNR improvement will not be as expected. This is because interpolation already used average of surrounding pixels and created correlation between them. Correlated pixels when binned won't produce expected SNR improvement like truly random noise samples.

[tim22]

Thank you both very much for those replies  (I wasn't quite sure where to put the post as although I am mainly interested in the live EEA aspect and flit from target to target I do also capture like to capture some of the images bit I am not a serious imager).  I use Sharpcap which allows   you to do both things while live stacking.

In answer to your first point Martin I  did wonder about the basis of the DSS score  but, like you, guess that it must be independent of binning -- but maybe 1.6 arcsec / pixel  was indeed too close to the seeing FWHM?  

Vlaiv, your point about the difference in the 1x and 2X noise floor and the way that DSS calculates FWHM makes a lot of sense.  Sharpcap gave similar FWHM pix estimates but it probably works in a similar way?    I hadn't thought of that at all . It is a significant point of difference between 1x and 2x binned images and maybe particularly so for such short exposures?    

I will have a look at Astroimage as you suggest.  That is also an interesting point that you make about whether or not binning was carried out before or after debayering/ pixel interpolation.  The software binning was carried out in Sharpcap and I expect that Robin Glover is aware of the potential problem that you mention but I will ask him.  By visual inspection I would say that the 2x binned images do show  significantly improved SNR  but are not quite as sharp as 1X.   

best wishes

Tim H

 

[vlaiv]

Tim,

I just came back from a mini session in my basement - received artificial star and went to asses quality of Samyang 85mm F/1.4 lens with my ASI1600. I used SharpCap to capture images just to compare star shapes in center of the frame and on the edge at different aperture settings - F/1.4, F/2, F/2.8 and F/4.

I planned to capture Lum, R, G, B and narrowband Ha and ~540nm Baader solar continuum (good for assessing quality of optic at near best visual wavelength - nice for refractors / achromats), but only manage to do Lum tonight.

In any case - I noticed that SharpCap behaves the same - I needed to change exposure length to compensate for aperture stop and used FWHM measurement for best focus at each setting. Changing of exposure length had immediate effect on measured FWHM value although it should make no difference. This of course has to do with noise floor as longer exposure will improve SNR.

Yes, SharpCap (at least version 2.9 that I have installed on my laptop) behaves like that - measured FWHM depends on SNR of the star - higher the SNR larger the FWHM.

[tim22]

Hi  Vlaiv,

That is an interesting set of experiments.  Firstly I like the idea of having a set up like that to test out optics.  It is a waste of precious clear sky time to be doing that in the field.  It nicely confirms what you said about the relationship between FWHM values measured in DSS /  Sharpcap and SNR.  Also- since it is a controlled 'lab' experiment -   it  takes out any potential  for a greater 'guiding'  contribution to FWHM at longer exposures.   

I suppose that means  that  the higher FWHM values  at bin2 are probably providing the more realistic indication of the seeing.  Anyway I think that you have solved the puzzle and have certainly taught me something.  Thanks again

Tim