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

Stacking DSO with Autostakkert - Who's tried it?

  • Please log in to reply
8 replies to this topic

#1 Borodog

Borodog

    Apollo

  • -----
  • topic starter
  • Posts: 1,338
  • Joined: 26 Oct 2020

Posted 21 January 2021 - 10:48 PM

I like to do stuff with stuff that is not meant for that stuff. One thing I'm going to try out tomorrow night is stacking DSO with Autostakkert. Nominally people use AS for planetary and lunar, but set the correct options and it seems capable of stacking deep sky images as well. Write out the results in FITS format and post-process in APP or what have you just like anything else. If you have not captured with flats and darks AS supports generating and using them as well, including dark flat or bias calibrated flats (although I had some trouble with my flats from a lunar mosaic last night; the flats did not calibrate out properly. I think something in the image train may have moved or something).

 

One of the reasons I am attracted to this idea is that you can save the frames in a single .ser format video file in SharpCap and AS does a pretty good job of sorting frames and makes them easy to review by quality. I know from lunar imaging that the tracking and stacking are top notch. Another reason that I am drawn to it is that frankly the stacks look like the lights. Perhaps this is silly, but I cannot get over how DSS, for example, produces a stack that looks absolutely nothing like the lights. Drives me nuts. For what it's worth APP doesn't suffer from this problem either.

 

So, has anybody used AS for stacking DSO images? If so, how did it go? Any deficiencies? I assume there must be some, otherwise it would be more popularly used for this, yes? Or has it just been made obsolete for this use, like Registax for anything but wavelet sharpening (sorry Registax stackers)? I know that altaz mount users might have problems from field rotation, but my mounts are all eq. 

 

 



#2 Stelios

Stelios

    Voyager 1

  • *****
  • Moderators
  • Posts: 10,857
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 22 January 2021 - 02:11 AM

You mean the stacks look like the *autostretched* lights. 

 

Individual light subs are (should be!) extremely dark, with a few bright stars showing at most (extremely bright DSO like Andromeda's core and Orion's Trapezium region excepted). But most acquisition software shows them auto-stretched. When DSS stacks them, it does *not* stretch, hence the "disappointment."

 

I don't see how AS!3 can generate flats and darks for you. The camera (and an acquisition program, Sharpcap in your case I believe) needs to do that, not software. 

 

I think your experiment will fail, but by all means go ahead and try it--you can always reprocess in DSS or APP or PI the same images if it doesn't work. The reason I think it will fail is that AS!3 doesn't by default align stars (to my knowledge), also doesn't use any of the traditional means of DSO image evaluation in its rankings (unlike, say, PI's Subframe Selector), so you won't know on what basis it will rank images. There may be other reasons, but again--interested to see what you can actually produce. The proof will be in the pudding, if any. 

 

To make the test meaningful, you should stack in both AS!3 and DSS or APP or PI, so you can compare the results. 


  • bobzeq25 and Borodog like this

#3 Tapio

Tapio

    Soyuz

  • -----
  • Posts: 3,864
  • Joined: 24 Sep 2006
  • Loc: Tampere, Finland

Posted 22 January 2021 - 02:52 AM

I have done some DSO stacking with Registax but not with Autostakkert.

Guess it will work but why do it when there's better tools for it.


  • Borodog likes this

#4 james7ca

james7ca

    Cosmos

  • *****
  • Posts: 9,244
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 22 January 2021 - 07:10 AM

Yes, you can stack DSOs with Autostakkert! I've done that several times. But, usually only when I've imaged small targets using so-called lucky imaging techniques that may require thousands of subs and which may only have a few potential alignment points.

 

In fact, if you want to see some really fine DSO images that were stacked in Autostakkert! just go to the website of the author of Autostakkert! (Emil Kraaikamp). Use the link below to see some of Emil's images:

 

  https://www.astrokra...?t=y&category=7


  • Borodog likes this

#5 Borodog

Borodog

    Apollo

  • -----
  • topic starter
  • Posts: 1,338
  • Joined: 26 Oct 2020

Posted 22 January 2021 - 09:36 AM

You mean the stacks look like the *autostretched* lights. 

 

Individual light subs are (should be!) extremely dark, with a few bright stars showing at most (extremely bright DSO like Andromeda's core and Orion's Trapezium region excepted). But most acquisition software shows them auto-stretched. When DSS stacks them, it does *not* stretch, hence the "disappointment."

 

I don't see how AS!3 can generate flats and darks for you. The camera (and an acquisition program, Sharpcap in your case I believe) needs to do that, not software. 

 

I think your experiment will fail, but by all means go ahead and try it--you can always reprocess in DSS or APP or PI the same images if it doesn't work. The reason I think it will fail is that AS!3 doesn't by default align stars (to my knowledge), also doesn't use any of the traditional means of DSO image evaluation in its rankings (unlike, say, PI's Subframe Selector), so you won't know on what basis it will rank images. There may be other reasons, but again--interested to see what you can actually produce. The proof will be in the pudding, if any. 

 

To make the test meaningful, you should stack in both AS!3 and DSS or APP or PI, so you can compare the results. 

No, that’s not what I mean at all, and is in fact the exact opposite. Autostakkert stacks look exactly like the lights, just without the noise and at higher bit depth. Autostakkert does no stretching at all. DSS stacks look absolutely nothing like the lights. 
 

And I meant that AS has tools to let you load dark frames and flat frames and produce master darks and flats and then apply them while stacking your lights.

 

I’m not sure why you think AS won’t track stars? It will track any feature at all in the image. 
 

I am unfamiliar with “the traditional means of DSO image evaluation.” This is the main area that I think AS might fall flat in. Namely, will AS’s quality evaluator be able to pick out tight round stars as better than slightly smeared or oblong stars? I’m not sure.

 

I will be testing AS against both DSS and APP on the same lights and calibration frames.

 

I have done some DSO stacking with Registax but not with Autostakkert.

Guess it will work but why do it when there's better tools for it.

Because it is simple and intuitive and produces stacks that make sense to me, and I am by now very familiar with it, and as I said, I find the idea of being able to stack ser files attractive. 

 

Yes, you can stack DSOs with Autostakkert! I've done that several times. But, usually only when I've imaged small targets using so-called lucky imaging techniques that may require thousands of subs and which may only have a few potential alignment points.

 

In fact, if you want to see some really fine DSO images that were stacked in Autostakkert! just go to the website of the author of Autostakkert! (Emil Kraaikamp). Use the link below to see some of Emil's images:

 

  https://www.astrokra...?t=y&category=7

Thank you for reminding me! I have seen those before and completely forgot.



#6 james7ca

james7ca

    Cosmos

  • *****
  • Posts: 9,244
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 22 January 2021 - 09:57 PM

One issue with AutoStakkert! (AS!3), it only does pixel rejection on mono images and it supports only one rejection algorithm (sigma clipping). That said, I consider pixel reject kind of a cosmetic operation as it doesn't really improve the signal to noise of your finished image. Thus, one shouldn't consider it a real noise reduction method and no matter what algorithm you use it should not make a huge difference in the overall appearance of your image (unless you have satellite or airplane tracks or something like that, but in that case the challenge of pixel rejection is to remove that artifact without having a detrimental effect on your overall signal to noise).

 

Plus, I don't believe that AS!3 does any weighting on the integration of the data, except by the user's own selection on the number of subs to combine (a quality estimation based upon some value of sharpness or detail). However, most (all?) stacking software that is designed for deep space objects combine on the basis of signal to noise, although a program like PixInsight lets you weight each sub on multiple factors.

 

Lastly, AS!3's registration most likely won't be as accurate as could be had with traditional stacking tools (on DSO targets).

 

Given all of the above, I think you'd have to rate AS!3 as a somewhat poor substitute for traditional DSO stacking software. However, it does excel in planetary stacking.



#7 Borodog

Borodog

    Apollo

  • -----
  • topic starter
  • Posts: 1,338
  • Joined: 26 Oct 2020

Posted 23 January 2021 - 06:45 AM

Thanks for the thoughts. My experiments will have to wait for another night as I wasn’t able to get any lights in last night.

#8 rj144

rj144

    Mariner 2

  • -----
  • Posts: 239
  • Joined: 31 Oct 2020

Posted 23 January 2021 - 02:49 PM

This is interesting.  I never thought about this, but maybe I could use this with Atl-Az mount for shorter/longer focal length exposures.

 

How does Autostakkert do the aligning though?  I would think if you used a typical frame rate, not many stars will be visible at all.



#9 james7ca

james7ca

    Cosmos

  • *****
  • Posts: 9,244
  • Joined: 21 May 2011
  • Loc: San Diego, CA

Posted 24 January 2021 - 04:20 AM

As long as there is at least one reasonably bright star (or object) that is a decent distance from the edge of the frame then AutoStakkert! should be able to align the frames. But, I don't think AutoStakkert! can handle any rotation or significant distortions between frames. This is one reason why a program like PixInsight (or even DSS) will probably produce better registration.




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