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

ASI 1600MM Pro Ha narrowband low SNR issue.

  • Please log in to reply
9 replies to this topic

#1 BoskoSLO

BoskoSLO

    Explorer 1

  • -----
  • topic starter
  • Posts: 61
  • Joined: 04 Jul 2018
  • Loc: North Slovenia

Posted 17 July 2019 - 11:07 AM

Today I tried to process my Narrowband data of the Elephant Trunk nebula from the last 2 nights and found 2 issues.

 

The first one is that dithering in RA only completely ruined my 3 hours worth of data because of diagonal banding across the whole image and I have no idea how to get rid of it (I might have to mix a lot more data with dithering in DEC as well to cover up the mistake but that will take some time).

 

The second issue is that I think even after stacking 3 hours worth of good frames (after using SubframeSelctor) I noticed that there is still a lot of noise in the stacked image. When I check other peoples images on astrobin using the same camera and with mostly the same gain and offset settings, their  Ha narrowband images looks much more sharp and with "more dynamic range?" (dark nebulas are dark and not littered with grey pixels).

 

The best way to explain what I'm talking about is to post some examples:

TS Photoline 560/80 x0.8

Orion Atlas

ASI 1600MM Pro   unity gain -15C

300'' subs with 12nm Ha Astronomik 2'' filter

 

This is the stacked 3,5h of data. Notice the diagonal banding shaped like "/"

Tz3Jk3C.jpg

 

 

This is the 1,2h stacked with dithering in RA and DEC. Notice there is 0 banding issue

MPc6ZE3.jpg

 

 

This is the 2.3h stacked with RA only dithering. You can see that there is lots of diagonal banding

 

EDIT: I just realized that the banding direction is on the RA axis because I didn't dithered in DEC the banding just stacked as signal. Makes sense now.

kqdH55L.jpg

 

side by side comparison

xw5kA8r.png

 

 

So the banding was easy to fix by enabling dithering in DEC even with high backlash but it worked well on the last session.

The main concern is the low SNR in the stacked image. I tried stacking without flats nor darks and the final image is pretty much identical.

I considered that maybe the 12nm Ha is not narrow enough and the full moon sneaks into my light frames and reducing SNR?

Or maybe I'm just stacking it wrong. I'm following these steps found here:

https://www.lightvor...t.html#Section4

 

I'm pretty much a newbie as far as mono cameras goes and their settings so maybe you can give me advice to improve.

 

Clear skies!


Edited by BoskoSLO, 17 July 2019 - 11:09 AM.


#2 2ghouls

2ghouls

    Viking 1

  • *****
  • Posts: 687
  • Joined: 19 Sep 2016

Posted 17 July 2019 - 11:23 AM

Great analysis regarding the dithering. I agree with your conclusions, and you should be able to avoid the banding by dithering in both axes going forward.

 

As far as low SNR compared to other people's images: 

You seem to be basing this visually off people's finished images on Astrobin. This isn't a great way to judge this because most people use noise reduction in post, especially in the shadow areas that you are talking about.  A better way might be to ask someone to share their data to compare, but even then local sky conditions are complex and I know from experience that just saying "red zone" or "white zone" doesn't make two skies directly comparable.

 

Okay, all that said. Looking at your 1.2 hour integration where you were dithering in both RA and Dec, I think the SNR looks good, and exactly what I would expect from this camera at unity gain at f/5.6. I don't think you have anything to worry about. The only thing I would suggest is to try gain 200. I shoot at f/5.4, and always use gain 200 when shooting narrowband. Usually 300" exposures. With those settings I only clip bright star cores, which I am fine with.



#3 james7ca

james7ca

    Fly Me to the Moon

  • *****
  • Posts: 7170
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 17 July 2019 - 11:25 AM

You might be able to remove some of the diagonal banding by rotating the frame to make the bands horizontal and then running the CanonBandingReduction script. Then rotate the frame back to its original orientation. This won't fix everything and the banding reduction procedure can introduce defects of its own making (i.e. the lesser of two evils), but it might be worth trying.

 

As for the signal-to-noise, as you might expect the usual solution to that problem is just more exposure time. Also, the Astrobin images that you are looking at have obviously been processed and so that makes comparisons difficult since some people are just better at removing or "hiding" noise in their finished images. As for the 12nm Ha filter, yes, that will suffer more under moonlit skies than something that is narrower.



#4 BoskoSLO

BoskoSLO

    Explorer 1

  • -----
  • topic starter
  • Posts: 61
  • Joined: 04 Jul 2018
  • Loc: North Slovenia

Posted 17 July 2019 - 11:32 AM

Great analysis regarding the dithering. I agree with your conclusions, and you should be able to avoid the banding by dithering in both axes going forward.

 

As far as low SNR compared to other people's images: 

You seem to be basing this visually off people's finished images on Astrobin. This isn't a great way to judge this because most people use noise reduction in post, especially in the shadow areas that you are talking about.  A better way might be to ask someone to share their data to compare, but even then local sky conditions are complex and I know from experience that just saying "red zone" or "white zone" doesn't make two skies directly comparable.

 

Okay, all that said. Looking at your 1.2 hour integration where you were dithering in both RA and Dec, I think the SNR looks good, and exactly what I would expect from this camera at unity gain at f/5.6. I don't think you have anything to worry about. The only thing I would suggest is to try gain 200. I shoot at f/5.4, and always use gain 200 when shooting narrowband. Usually 300" exposures. With those settings I only clip bright star cores, which I am fine with.

Yeah I have been reading a lot regarding gain settings for the ASI 1600MM Pro and got mixed answers. Everybody has a slightly different opinion which gain to use and why. Unity gain is like a "jack of all trades" and as far as LRGB imaging goes it works great :

 

check my astrobin "BoskoSLO" and my LRGB Elephant trunk nebula image (For some reason CloudyNights doesn't allow link in comments).

https://www.astrobin...2769/0/?nc=user

 

So yeah I will try gain 200 on the next clear night and maybe even max gain 300.


Edited by BoskoSLO, 17 July 2019 - 11:33 AM.


#5 BoskoSLO

BoskoSLO

    Explorer 1

  • -----
  • topic starter
  • Posts: 61
  • Joined: 04 Jul 2018
  • Loc: North Slovenia

Posted 17 July 2019 - 11:46 AM

You might be able to remove some of the diagonal banding by rotating the frame to make the bands horizontal and then running the CanonBandingReduction script. Then rotate the frame back to its original orientation. This won't fix everything and the banding reduction procedure can introduce defects of its own making (i.e. the lesser of two evils), but it might be worth trying.

 

As for the signal-to-noise, as you might expect the usual solution to that problem is just more exposure time. Also, the Astrobin images that you are looking at have obviously been processed and so that makes comparisons difficult since some people are just better at removing or "hiding" noise in their finished images. As for the 12nm Ha filter, yes, that will suffer more under moonlit skies than something that is narrower.

Tried it and the script doesn't like rotated images. To salvage the frames I just need as you said do more imaging with RA and DEC dither with maybe higher gain and I might need to replace the filter with a 5 or 6nm version at some point. And at the end of the day I wouldn't have managed to get so much signal out of this nebula with my modified Canon 600D anyway.



#6 2ghouls

2ghouls

    Viking 1

  • *****
  • Posts: 687
  • Joined: 19 Sep 2016

Posted 17 July 2019 - 12:08 PM

Yeah I have been reading a lot regarding gain settings for the ASI 1600MM Pro and got mixed answers. 

Honestly, it doesn't make a huge difference (139 vs. 200), and I agree that gain 139 is a good middle ground to pick for everything. After some experimentation early on with the following gains: 0, 75, 139, 200, 300, I've settled on gain 200 for narrowband and gain 75 for broadband. Typical exposure lengths are: 90-180" for broadband and 180-600" for narrowband. At the longer end of those exposure ranges, I am definitely clipping lots of stars, so for most objects I am at 120" for broadband and 300" for narrowband. 

 

Nice photo!



#7 BoskoSLO

BoskoSLO

    Explorer 1

  • -----
  • topic starter
  • Posts: 61
  • Joined: 04 Jul 2018
  • Loc: North Slovenia

Posted 17 July 2019 - 12:16 PM

Honestly, it doesn't make a huge difference (139 vs. 200), and I agree that gain 139 is a good middle ground to pick for everything. After some experimentation early on with the following gains: 0, 75, 139, 200, 300, I've settled on gain 200 for narrowband and gain 75 for broadband. Typical exposure lengths are: 90-180" for broadband and 180-600" for narrowband. At the longer end of those exposure ranges, I am definitely clipping lots of stars, so for most objects I am at 120" for broadband and 300" for narrowband. 

 

Nice photo!

The problem I have is that I pixel peek too much. The details that don't matter that much when you look at a picture from a distance away like a PC wallpaper or a printed image that is framed on a wall. I really need to stop pixel peeking for every single insignificant error that messes with my head. Just like it did with the PHD2 guiding graph. I was so obsessed with it but finally realized that all of my frames have perfect stars.

I managed to bring down my RMS error to 0.6 to 0.9'' by just installing a different driver for my guide camera, the GP mono CAM from altair 1.2MPixel. Maybe it's just a placebo effect because before my guiding was usually in the 1,4''RMS range for most of my imaging sessions but for the past 2 days it's been half that for a total of 7 hours which is promising.


  • bmhjr likes this

#8 gatsbyiv

gatsbyiv

    Viking 1

  • *****
  • Posts: 661
  • Joined: 29 Mar 2009
  • Loc: Doylestown, PA

Posted 17 July 2019 - 12:21 PM

Most final images you see have had a good bit of careful contrast enhancement done to them (LHE, Curves, MMT, etc), so don't expect to see that in the initial integration.  What you've got here looks like it should, especially given the 12 nm bandpass and the presence of the moon.  The fastest way to improved contrast would be more integration time and a 5 or 6 (or 3!) nm filter. 

 

(And as you have probably already found, cropping out the black edges from your image allows ScreenTransferFunction to do a better stretch.  Running DynamicBackgroundExtraction also usually allows you to do a better stretch, but that's difficult for an image like this, where the object fills the frame.)



#9 WalterG

WalterG

    Sputnik

  • -----
  • Posts: 40
  • Joined: 26 Dec 2011

Posted 17 July 2019 - 02:26 PM

Interesting!  I have had the same banding issues with my ASI and I never thought of dithering as the problem.  Thanks for the insight.

Walter 



#10 BoskoSLO

BoskoSLO

    Explorer 1

  • -----
  • topic starter
  • Posts: 61
  • Joined: 04 Jul 2018
  • Loc: North Slovenia

Posted 17 July 2019 - 02:45 PM

Interesting!  I have had the same banding issues with my ASI and I never thought of dithering as the problem.  Thanks for the insight.

Walter 

yeah in my instance proper dither makes such a big difference. my Image scale is 1,75''/pixel and I dither anywhere from 10 to 20 arc seconds in random directions from the center if the guide star.

dithering in RA works really well but not for the final image as you can see. I have my guiding enabled in one direction only and sometimes it dithers in the wrong direction and PHD2 fails to bring back the star fully to the center so the DEC floats ~4 arc seconds away from the center but as song as PA is good enough the star wont drift far enough for the 5 minute frame to be ruined until the next dither event when it usually fixes it self.

Anyway here is the result of me eye balling the post processing in PixInSight. Making the dark nebula darker is easier than in RGB image I realized. I think, I underestimated the data I gathered and thought it's bad. I will mix it with my LRGB data later this week smile.gif

 

GzoKWyI.jpg


Edited by BoskoSLO, 17 July 2019 - 02:45 PM.

  • WalterG and tkottary like this


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