Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

What FWHM to aim for?

  • Please log in to reply
30 replies to this topic

#1 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 10:43 AM

Hi everyone, as I start to try to tune up my autoguiding I'm trying to figure out what a "good" FWHM value would be for my setup.

 

I'm shooting with a Canon T5i on a William Optics 66SD which yields 2.29"/pixel. With 120 second subs of the elephant trunk last night I was getting between 3.83 - 4.00 FWHM values recorded in DSS (in pixels I believe). Conditions were pretty decent last night, and these values seem pretty typical of my setup so far.

 

That doesn't seem good. The stars are nice and round, and I've been able to product images with decent detail, but I do wonder how good I might be able to get if I tune things up a bit. 

 

So, anyone have thoughts on what a good goal would be?



#2 Dynan

Dynan

    Fly Me to the Moon

  • *****
  • Posts: 5,991
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 23 December 2018 - 11:02 AM

Lowest possible, which will vary nightly according to seeing. If you see a wide variance while focusing, it's probably seeing conditions.

 

Are you using BYEOS? Guylain explains FWHM very well.

 

https://www.youtube....h?v=z3gkw8bx7Aw  (at about 38:00)

 

I believe it's just a ratio of width to calculated height, but I may be wrong.



#3 terry59

terry59

    Voyager 1

  • *****
  • Posts: 10,457
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 23 December 2018 - 11:16 AM

Ummm...how do you focus? What do you measure FWHM with?



#4 CharlesW

CharlesW

    Long time member

  • ***--
  • Posts: 3,659
  • Joined: 01 Nov 2012
  • Loc: Chula Vista, CA

Posted 23 December 2018 - 11:24 AM

If you are manually focusing I would not waste my time chasing a constantly varying FWHM number in any app. Get and use a B-Mask. You’ll have better results, and faster. After that you’ll have to figure out how much wind is too much. I’m down to gusts of 10 mph and I’m open to less than that. If where you live is cloudy a lot, you may need to be a little more forgiving. 



#5 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 11:34 AM

To be clear, I'm talking about the FHWM numbers reported by DSS on long exposures, not those reported during focusing. I do use a bahtinov mask for focusing. I think I'm doing relatively well there, but always room for improvement I'm sure. 

 

My concern at the moment is figuring out how much my tracking is hurting (or helping I guess). I know I've got room to improve my autoguiding. I just don't know how good I can reasonably expect my setup to achieve, and hence, when to be satisfied. 

 

If I do the math on my current results: 3.83 pixel FWHM * 2.29"/pixel = 8.77 arcseconds resolution. That doesn't sound good.



#6 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,965
  • Joined: 27 Oct 2014

Posted 23 December 2018 - 12:10 PM

To be clear, I'm talking about the FHWM numbers reported by DSS on long exposures, not those reported during focusing. I do use a bahtinov mask for focusing. I think I'm doing relatively well there, but always room for improvement I'm sure. 

 

My concern at the moment is figuring out how much my tracking is hurting (or helping I guess). I know I've got room to improve my autoguiding. I just don't know how good I can reasonably expect my setup to achieve, and hence, when to be satisfied. 

 

If I do the math on my current results: 3.83 pixel FWHM * 2.29"/pixel = 8.77 arcseconds resolution. That doesn't sound good.

It's not, unfortunately.

 

But the FWHM numbers others get have little relevance to your situation.  Way too many variables.  FWHM is mostly something you check on a relative basis for yourself, to get a feel for how things are going and how good the night is where you are.

 

That said, here's a rough rule of thumb for the average imager that may be useful in your situation.  One significant figure is appropriate.  Most people should be able to get below 5 arc sec on most nights.  (Seeing can always do you in.)  Below 4 is good, below 3 excellent.  Below 2 is advanced imager in great skies territory.


Edited by bobzeq25, 23 December 2018 - 12:15 PM.

  • pgs/sdg and Dynan like this

#7 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 02:02 PM

It's not, unfortunately.

 

But the FWHM numbers others get have little relevance to your situation.  Way too many variables.  FWHM is mostly something you check on a relative basis for yourself, to get a feel for how things are going and how good the night is where you are.

 

That said, here's a rough rule of thumb for the average imager that may be useful in your situation.  One significant figure is appropriate.  Most people should be able to get below 5 arc sec on most nights.  (Seeing can always do you in.)  Below 4 is good, below 3 excellent.  Below 2 is advanced imager in great skies territory.

Thanks Bob, I appreciate the feedback. If nothing else, just being told that I can do better is helpful. 

 

Interestingly, I just restacked that same set of data in APP, and it is reporting completely different FWHM values. It says my best image ranges from 2.60 - 2.85 FWHM. Take the high side of that and now we're talking 6.5 arc seconds. Better, but sounds like still ought to be able to do better. Mind you, I have no idea which numbers to believe...

 

Next up is trying to use PEC on my mount to see if that makes PHD's job easier. That's assuming Santa brings me clear skies for Xmas.



#8 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,929
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 23 December 2018 - 02:37 PM

Hi everyone, as I start to try to tune up my autoguiding I'm trying to figure out what a "good" FWHM value would be for my setup.

 

I'm shooting with a Canon T5i on a William Optics 66SD which yields 2.29"/pixel. With 120 second subs of the elephant trunk last night I was getting between 3.83 - 4.00 FWHM values recorded in DSS (in pixels I believe). Conditions were pretty decent last night, and these values seem pretty typical of my setup so far.

 

That doesn't seem good. The stars are nice and round, and I've been able to product images with decent detail, but I do wonder how good I might be able to get if I tune things up a bit. 

 

So, anyone have thoughts on what a good goal would be?

If you are using an unmodified DSLR, then my guess is the filter stack over the sensor is probably one of the key reasons your FWHMs are large. Most terrestrial cameras have a low pass filter over the sensor, and its express purpose is to blur high frequency data to avoid issues with aliasing (notably, the moire artifact that can occur with repeating patterns in subjects.) Even if your seeing was good, the low pass filter alone could still be causing your FWHMs to bloat to ~3-4 pixels...would depend on how strong the filter is in the T5i.

 

DSLRs are also color cameras (with a very few exceptions), which means they must be demosaiced to produce a full color image. That also introduces some blur.

 

Two other key factors will affect your FWHMs. The size of your scope...smaller apertures will increase the amount of diffraction, and that can account for 3-4" (not pixels) of FWHM in and of itself. The final factor is focus. At ideal focus over 70% of the energy of the star will be in the central peak, but as you lose focus a lot of that energy will be thrown out into the halo of the star rather quickly. 

 

It is most likely that your seeing is not 9" (that would be atrocious, and you would most definitely know if you looked at the stars in the sky yourself), probably not even 3" if you felt the night was good. I would say the top two factors are the low pass filter and focus, with diffraction coming in a close second. Seeing unless the skies looked visibly bad, could be the least significant factor here (3" seeing is pretty bad, 4" is terrible...I myself will stop imaging if seeing gets to 4", and I often toss subs where seeing is over 3".)


  • Michael Covington, pgs/sdg and Dynan like this

#9 terry59

terry59

    Voyager 1

  • *****
  • Posts: 10,457
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 23 December 2018 - 02:38 PM

Thanks Bob, I appreciate the feedback. If nothing else, just being told that I can do better is helpful. 

 

Interestingly, I just restacked that same set of data in APP, and it is reporting completely different FWHM values. It says my best image ranges from 2.60 - 2.85 FWHM. Take the high side of that and now we're talking 6.5 arc seconds. Better, but sounds like still ought to be able to do better. Mind you, I have no idea which numbers to believe...

 

Next up is trying to use PEC on my mount to see if that makes PHD's job easier. That's assuming Santa brings me clear skies for Xmas.

Using my 80ED doublet here in Colorado with the attendant seeing we have, my FWHM was around 2.2 - 2.3


  • Scott Mitchell likes this

#10 vio

vio

    Mariner 2

  • -----
  • Posts: 269
  • Joined: 27 Apr 2011
  • Loc: Naples, FL

Posted 23 December 2018 - 04:08 PM

Depending on the stacking method used, and pixel rejection rules, the FWHM of a frame can be higher.


Sent from my iPad using Tapatalk

#11 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 04:35 PM

If you are using an unmodified DSLR, then my guess is the filter stack over the sensor is probably one of the key reasons your FWHMs are large. Most terrestrial cameras have a low pass filter over the sensor, and its express purpose is to blur high frequency data to avoid issues with aliasing (notably, the moire artifact that can occur with repeating patterns in subjects.) Even if your seeing was good, the low pass filter alone could still be causing your FWHMs to bloat to ~3-4 pixels...would depend on how strong the filter is in the T5i.

 

DSLRs are also color cameras (with a very few exceptions), which means they must be demosaiced to produce a full color image. That also introduces some blur.

 

Two other key factors will affect your FWHMs. The size of your scope...smaller apertures will increase the amount of diffraction, and that can account for 3-4" (not pixels) of FWHM in and of itself. The final factor is focus. At ideal focus over 70% of the energy of the star will be in the central peak, but as you lose focus a lot of that energy will be thrown out into the halo of the star rather quickly. 

 

It is most likely that your seeing is not 9" (that would be atrocious, and you would most definitely know if you looked at the stars in the sky yourself), probably not even 3" if you felt the night was good. I would say the top two factors are the low pass filter and focus, with diffraction coming in a close second. Seeing unless the skies looked visibly bad, could be the least significant factor here (3" seeing is pretty bad, 4" is terrible...I myself will stop imaging if seeing gets to 4", and I often toss subs where seeing is over 3".)

Interesting point about the filters inherent in the DSLR. Mine is modded but not a full spectrum mod. So certainly at least one filter still in play. Actually, there's also my CLS filter which is intentionally in the imaging train. Hopefully it doesn't cause degradation. 

 

And yeah, I'm sure seeing isn't the root of my high FWHM values. It is a contributor for sure, but I'm also sure my guiding isn't dialed in yet. There may be another post soon to help diagnose that if I can't get it to improve on the next couple of nights out.

 

Thanks for the info!



#12 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 04:37 PM

Depending on the stacking method used, and pixel rejection rules, the FWHM of a frame can be higher.


Sent from my iPad using Tapatalk

Ah, good point vio! Yes, that probably explains the difference between APP and DSS. Also probably explains why I like the output of APP better as well. Now if it just didn't take 5x longer than DSS...



#13 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 24,965
  • Joined: 27 Oct 2014

Posted 23 December 2018 - 04:56 PM

Ah, good point vio! Yes, that probably explains the difference between APP and DSS. Also probably explains why I like the output of APP better as well. Now if it just didn't take 5x longer than DSS...

Chuckle.

 

Capabilities of processing programs are a direct function of how much time is involved.  PixInsight is arguably the best (a majority of advanced imagers think so).  Some people are put off by the $250 cost, but that's completely unimportant.  The real cost of PI is how much time it takes to learn it and use it.  I have hundreds of hours in that, so the cost of the program is less that $1/hour.  My time is considerably more valuable.

 

One minor thing you can do in PI, relevant to the thread.  You can measure your FWHM in each sub, look at graphs of the data.  Can set a limit of "FWHM must be less the 4" and PI will select out those subs and move them to an "approved" folder.  You can make the selection criteria _much_ more complicated.


  • Scott Mitchell likes this

#14 Michael Covington

Michael Covington

    Author

  • *****
  • Posts: 6,868
  • Joined: 13 May 2014
  • Loc: Athens, Georgia, USA

Posted 23 December 2018 - 04:57 PM

I usually say 3 pixels or 3 arc-seconds, whichever is larger.  As has been pointed out, most DSLRs don't expose the pixels directly to the image source -- they have a low-pass filter so that every point in the image will be spread over a few pixels and the Bayer matrix will work as intended.

With DSLRs, FWHMs below 3 pixels are uncommon.

There is also the limited resolution of your optics.



#15 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 05:21 PM

Ah, good point vio! Yes, that probably explains the difference between APP and DSS. Also probably explains why I like the output of APP better as well. Now if it just didn't take 5x longer than DSS...

Wait, check that. I'm conflating two different things here. The FWHM values I've been sharing are from the sub-exposures so the rejection algorithms aren't relevant here. However, APP does claim that's its Adaptive Airy Disk is a significant improvement when debayering so maybe that's actually what's making the difference in the reported FWHM values. Of course, that's also assuming that there's any consistency between DSS and APP on how the compute the average FWHM value.



#16 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 05:23 PM

I usually say 3 pixels or 3 arc-seconds, whichever is larger.  As has been pointed out, most DSLRs don't expose the pixels directly to the image source -- they have a low-pass filter so that every point in the image will be spread over a few pixels and the Bayer matrix will work as intended.

With DSLRs, FWHMs below 3 pixels are uncommon.

There is also the limited resolution of your optics.

That's interesting Michael, thanks. In that case I'd be looking at 3 pixels for my configuration, and that I'm in the ballpark with DSS's numbers and doing great according to APP's numbers. 


  • Michael Covington likes this

#17 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 05:42 PM

Chuckle.

 

Capabilities of processing programs are a direct function of how much time is involved.  PixInsight is arguably the best (a majority of advanced imagers think so).  Some people are put off by the $250 cost, but that's completely unimportant.  The real cost of PI is how much time it takes to learn it and use it.  I have hundreds of hours in that, so the cost of the program is less that $1/hour.  My time is considerably more valuable.

 

One minor thing you can do in PI, relevant to the thread.  You can measure your FWHM in each sub, look at graphs of the data.  Can set a limit of "FWHM must be less the 4" and PI will select out those subs and move them to an "approved" folder.  You can make the selection criteria _much_ more complicated.

Bob, you don't happen to be a Robert Heinlein fan do you? It all comes back to tanstaafl!


  • bobzeq25 likes this

#18 vio

vio

    Mariner 2

  • -----
  • Posts: 269
  • Joined: 27 Apr 2011
  • Loc: Naples, FL

Posted 23 December 2018 - 05:48 PM

Wait, check that. I'm conflating two different things here. The FWHM values I've been sharing are from the sub-exposures so the rejection algorithms aren't relevant here. However, APP does claim that's its Adaptive Airy Disk is a significant improvement when debayering so maybe that's actually what's making the difference in the reported FWHM values. Of course, that's also assuming that there's any consistency between DSS and APP on how the compute the average FWHM value.


Sorry I misinterpreted what you said about better numbers reported after stacking, now I realize this was on stats from frames, not resulting image.


Sent from my iPad using Tapatalk
  • Scott Mitchell likes this

#19 the Elf

the Elf

    Soyuz

  • *****
  • Posts: 3,686
  • Joined: 06 Sep 2017
  • Loc: Germany

Posted 23 December 2018 - 07:21 PM

More or less confirming Bob's data. To be precise: I use PI to rate the images. First step I platesolve (there is a solver in PI) to get the exact arcsec/pixel. I calibrate all subs (bias/dark/flat frames), in case of OSC I debayer and feed it into the subframe selector script, entering the scale and display the results in arcsecs.  I toss everything above 5, I'm not frustrated between 3.5 and 4.5 and I get excited if it is less then 3.5. My best ever was 2.9. Important: I have the same values in arcsecs for my 1000+mm RC and my 420mm refractor. In addition I check images with the FWHMEccentricity script. As expected the refractor is flat like a pancake while the RC comes with serious curvature. The SSS values are average values. That means for the RC it is better in the center and far worse in the corners. Values are higher in general with the color DSLR as it contains a blur filter in front of the bayer pattern. The mono is stripped down to the sensor and comes with better values all the time. So, expect better values for mono. I followed a pro's conversation. One was complaining to use subs up to 2.5 though he would normally toss everything greater than 2. I guess that was a bad day on Mauna Kea.

After stacking deconvolution my reduce the value about 1 arcsec but the denoise adds 0.5 to 1 so my goal is to have the same FWHM in the linear stack after deconvolution and denoise. Make sure you do not bloat stars by denoise or stretching.

Don't forget that the smallest amount is 1 pixel. If you image at 5 arcsec/pixel 5 arcsec is perfect! Also think about the presentation. If you have a DSLR image (5000x4000 pixels) with a star diameter of 2 pixels and present the full field of view on your fullHD TV set, what have you got then? A single pixel. If you crop you may have two.


Edited by the Elf, 23 December 2018 - 07:27 PM.

  • Scott Mitchell, bobzeq25 and Boeglewatcher like this

#20 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 08:38 PM

If you are using an unmodified DSLR, then my guess is the filter stack over the sensor is probably one of the key reasons your FWHMs are large. Most terrestrial cameras have a low pass filter over the sensor, and its express purpose is to blur high frequency data to avoid issues with aliasing (notably, the moire artifact that can occur with repeating patterns in subjects.) Even if your seeing was good, the low pass filter alone could still be causing your FWHMs to bloat to ~3-4 pixels...would depend on how strong the filter is in the T5i.

Jon, one more question on that. I presume these filters are not a factor in color astro CMS cameras like the ASI071MC, or is that part of the design of the chip itself? I'm guessing that they are not in play and hence could expect better results with the purpose built camera (beyond the obvious advantages provided by cooling).



#21 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,929
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 23 December 2018 - 09:27 PM

Jon, one more question on that. I presume these filters are not a factor in color astro CMS cameras like the ASI071MC, or is that part of the design of the chip itself? I'm guessing that they are not in play and hence could expect better results with the purpose built camera (beyond the obvious advantages provided by cooling).

They are not used on astro cams....the ASI071 just has cover glass, but no LPF.


  • Scott Mitchell likes this

#22 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 09:34 PM

They are not used on astro cams....the ASI071 just has cover glass, but no LPF.

Perfect, thanks for confirming. It is an advantage of dedicated astro cams that I hadn't thought about before. Is it too late to change my Christmas list? wink.gif



#23 Michael Covington

Michael Covington

    Author

  • *****
  • Posts: 6,868
  • Joined: 13 May 2014
  • Loc: Athens, Georgia, USA

Posted 23 December 2018 - 10:14 PM

Some DSLRs lack the low-pass filter.  The Nikon D5300 is reportedly an example.

 

But let's back up a moment.  Is our goal to get low FWHM measurements or to get good pictures?  That filter is there (on most DSLRs) because it improves the picture, improving the performance of the Bayer matrix.



#24 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,929
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 23 December 2018 - 10:21 PM

Some DSLRs lack the low-pass filter.  The Nikon D5300 is reportedly an example.

 

But let's back up a moment.  Is our goal to get low FWHM measurements or to get good pictures?  That filter is there (on most DSLRs) because it improves the picture, improving the performance of the Bayer matrix.

It is there to avoid aliasing. That is not an issue with deep space images, so it is effectively pointless for what the OP is trying to do. 

 

There are some DSLRs that lack the filter entirely, but others actually "reverse" the filter with the second LPF in the stack, which still causes some blurring. 

 

If you want the sharpest images possible...well, mono is best. But an OSC astro cam without any filter over the sensor combined with bayer drizzling to integrate will deliver much sharper results. 



#25 Scott Mitchell

Scott Mitchell

    Apollo

  • *****
  • topic starter
  • Posts: 1,454
  • Joined: 28 Jul 2007
  • Loc: Atlanta-ish, GA

Posted 23 December 2018 - 10:25 PM

Some DSLRs lack the low-pass filter.  The Nikon D5300 is reportedly an example.

 

But let's back up a moment.  Is our goal to get low FWHM measurements or to get good pictures?  That filter is there (on most DSLRs) because it improves the picture, improving the performance of the Bayer matrix.

Well, my goal is to figure out how to maximize what I can do with the equipment I have. However, at some point I'm also going to upgrade and the more points I can make to my wife about why it's a good idea, the better. smile.gif




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics