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

short exposures

  • Please log in to reply
29 replies to this topic

#1 zoltrix

zoltrix

    Vostok 1

  • -----
  • topic starter
  • Posts: 102
  • Joined: 23 Feb 2021

Posted 22 July 2021 - 06:30 AM

Hello

Probably this question has been asked several times
Is it true that,a long,let's say,10 minutes exposure is equivalent to 20, 30 seconds short exposures, thanks to the modern CMOS cameras with low read noise ?
Are significant comparative tests, all things being equal, available ?

#2 pyrasanth

pyrasanth

    Aurora

  • *****
  • Posts: 4,606
  • Joined: 08 Jan 2016

Posted 22 July 2021 - 06:40 AM

All things being equal then mostly yes. However things, sadly are not equal, Longer exposures are prone to guiding errors, over saturated stars that lose colour and a build up of unwanted background noise.

 

I find with CMOS I quickly get into the laws of diminishing returns with longer exposures- I see the same signal but a lot more background light pollution.

 

Short subs that reach above the noise ceiling I have found is the way to go.....and I'd sooner lose a 30 second sub due to clouds than a 5 minute sub.


  • lee14, petert913 and ks__observer like this

#3 Dynan

Dynan

    Fly Me to the Moon

  • *****
  • Posts: 6,078
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 22 July 2021 - 07:14 AM

As a wise astronomer, (whose initials are Bobzeq), taught me...It's total photons collected that matters, as long as exposures are long enough to overcome sensor noise.


Edited by Dynan, 22 July 2021 - 07:28 AM.


#4 jiberjaber

jiberjaber

    Lift Off

  • -----
  • Posts: 2
  • Joined: 13 May 2020

Posted 22 July 2021 - 07:16 AM

All things being equal then mostly yes. However things, sadly are not equal, Longer exposures are prone to guiding errors, over saturated stars that lose colour and a build up of unwanted background noise.

 

I find with CMOS I quickly get into the laws of diminishing returns with longer exposures- I see the same signal but a lot more background light pollution.

 

Short subs that reach above the noise ceiling I have found is the way to go.....and I'd sooner lose a 30 second sub due to clouds than a 5 minute sub.

I agree - but don't forget the diminishing return of how long it takes to pre-process 20 images compared to 1 image, especially when you build up a lot of data  - there's certainly a sweet spot but it will also depend on how well it all goes in capturing (favoring good conditions, equipment etc) to how long needed to spend in front of the computer (with shorter subs)... I'm not sure there's a formula for that sweet spot though :) 



#5 daveco2

daveco2

    Mariner 2

  • *****
  • Posts: 248
  • Joined: 29 Aug 2016
  • Loc: Torrance, CA

Posted 22 July 2021 - 07:28 AM

At my Bortle8 site, long exposures get me good saturated images of light pollution.


  • pyrasanth, charlieb123, CowTipton and 2 others like this

#6 John Miele

John Miele

    Aurora

  • *****
  • Posts: 4,770
  • Joined: 29 May 2005
  • Loc: North Alabama

Posted 22 July 2021 - 07:29 AM

I agree - but don't forget the diminishing return of how long it takes to pre-process 20 images compared to 1 image, especially when you build up a lot of data  - there's certainly a sweet spot but it will also depend on how well it all goes in capturing (favoring good conditions, equipment etc) to how long needed to spend in front of the computer (with shorter subs)... I'm not sure there's a formula for that sweet spot though smile.gif

So true!

 

I am definitely a convert to shorter subs with my ASI1600MM pro camera. I rarely do more than 30 seconds with luminance at home. But I also need a lot of total time to overcome my local LP and build up a good S/N ratio. For galaxies, I try for a minimum of 6 hours. Well that works out to 720 subs to calibrate and stack. You can really start to choke up a laptop stacking that many subs. I ended up buying a more powerful desktop with a Ryzen 7 chip and 32GB of RAM just to help. And I'm already ready to upgrade to 64 GB of RAM! 

 

So short subs are great for all the reasons mentioned but beware of the PC workload!



#7 Mert

Mert

    Cosmos

  • *****
  • Posts: 8,268
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 22 July 2021 - 08:12 AM

Right about cpu time! Using my laptop ( i7 4 core ) it takes a long time to preprocess and have the stacked image.
Normally I don't go longer then 120 seconds per frame and likely that is way too much when imaging from my LP balcony.
Processing 500 frames takes my laptop some 5 hours at least, PI preprocessing apart!
Have to think about getting a desktop with a good Ryzen

#8 pyrasanth

pyrasanth

    Aurora

  • *****
  • Posts: 4,606
  • Joined: 08 Jan 2016

Posted 22 July 2021 - 08:36 AM

Right about cpu time! Using my laptop ( i7 4 core ) it takes a long time to preprocess and have the stacked image.
Normally I don't go longer then 120 seconds per frame and likely that is way too much when imaging from my LP balcony.
Processing 500 frames takes my laptop some 5 hours at least, PI preprocessing apart!
Have to think about getting a desktop with a good Ryzen

Hi Mert,

 

I use a Ryzen 3950 with 128 GB of RAM. It tears through a stack of 500 subs (depending on binning) in only a few minutes- get some fast storage as well and the time worry will be dispelled lol.gif


  • psandelle likes this

#9 Mert

Mert

    Cosmos

  • *****
  • Posts: 8,268
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 22 July 2021 - 08:44 AM

Hi Mert,

I use a Ryzen 3950 with 128 GB of RAM. It tears through a stack of 500 subs (depending on binning) in only a few minutes- get some fast storage as well and the time worry will be dispelled lol.gif

Hi Mark,

Yes, that would be a nice cpu, together with some powerful NVidia card to use GPU paralel processing.
It's high on my wishlist :shrug:

#10 pyrasanth

pyrasanth

    Aurora

  • *****
  • Posts: 4,606
  • Joined: 08 Jan 2016

Posted 22 July 2021 - 08:50 AM

Hi Mark,

Yes, that would be a nice cpu, together with some powerful NVidia card to use GPU paralel processing.
It's high on my wishlist shrug.gif

Good luck on getting an Nvidia card- there does not appear to be many about- its all those bit coin minersfrown.gif



#11 zoltrix

zoltrix

    Vostok 1

  • -----
  • topic starter
  • Posts: 102
  • Joined: 23 Feb 2021

Posted 22 July 2021 - 09:00 AM

WOW !

I did not expect to have such an answer
I supposed that short exposures were fairly well but the state of art of astro imaging were still
the long exposures method
ok, the longer post processing may be annoying bu why should I purchase a 30 kilos / 3000 usd mount then ?
Not to mention the auto guiding

#12 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 25,653
  • Joined: 27 Oct 2014

Posted 22 July 2021 - 09:44 AM

WOW !

I did not expect to have such an answer
I supposed that short exposures were fairly well but the state of art of astro imaging were still
the long exposures method
ok, the longer post processing may be annoying bu why should I purchase a 30 kilos / 3000 usd mount then ?
Not to mention the auto guiding

Like most of DSO AP this gets complicated.  Moreso than can be explained in a short post here.  Some basic points.

 

"Optimal" subexposure length depends sensitively on your light pollution level (more LP = shorter exposures) and optical speed (faster optics = shorter exposures).     Camera read noise plays a role.

 

Narrowband filters are a whole different ballgame, require rather long subs, and other issues come into play.

 

Basically you're trying to get somewhere between too many too short exposures (too much read noise) and too few too long exposures (saturated stars with no color, clipped highlights).  This video is a good place to start.

 

https://www.youtube....h?v=3RH93UvP358

 

Take home messages.  What others use for subexposure has little relevance for you.  Get subexposure "in the ballpark" and total imaging time is far more important.  Total photons is what counts, not how you divide the total into subs.

 

A good mount remains important.  Autoguiding is just about essential, unless you have a really expensive mount.  >$5000.  And maybe still a good idea then.

 

Bad news.  This is complicated.  Good news, you will never ever run out of new things to learn.

 

Guiding principle.  When in doubt follow the crowd.  A lot of smart people have been doing this for a long time, the collective wisdom is large.


Edited by bobzeq25, 22 July 2021 - 09:47 AM.

  • mtc, Dynan, David Boulanger and 1 other like this

#13 John Miele

John Miele

    Aurora

  • *****
  • Posts: 4,770
  • Joined: 29 May 2005
  • Loc: North Alabama

Posted 22 July 2021 - 09:48 AM

Hi Mert,

 

I use a Ryzen 3950 with 128 GB of RAM. It tears through a stack of 500 subs (depending on binning) in only a few minutes- get some fast storage as well and the time worry will be dispelled lol.gif

Wow! 128 GB of RAM...you are 'rockin it my friend...lol!

 

Maybe I should set my sights on a 128 GB upgrade instead of 64!


  • pyrasanth likes this

#14 Mert

Mert

    Cosmos

  • *****
  • Posts: 8,268
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 22 July 2021 - 10:28 AM

I've done some trials with short exposures, less then 0,5 seconds per sub. The FWHM goes down, and guiding becomes less critical.
Seeing becomes less critical as well!!
CPU has to work a lot more though, but you can get more detail/smaller detail.

#15 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 25,653
  • Joined: 27 Oct 2014

Posted 22 July 2021 - 11:38 AM

I've done some trials with short exposures, less then 0,5 seconds per sub. The FWHM goes down, and guiding becomes less critical.
Seeing becomes less critical as well!!
CPU has to work a lot more though, but you can get more detail/smaller detail.

That can work, some.  You'll lose some dim detail in read noise from many subs, which shows up as background noise that swamps dim detail.

 

How much you lose depends on the usual suspects, the read noise spec of the camera, your light pollution level, your optical speed.

 

I got some decent images on a recent dark sky trip, with 8 second subs.  400mm F2 RASA, 2600MC camera with low read noise.  PhD2 was acting up, so I shot unguided on a CEM60, carefully polar aligned.

 

https://www.cloudyni...-1-part-1-of-3/


Edited by bobzeq25, 22 July 2021 - 11:40 AM.


#16 stargzr66207

stargzr66207

    Viking 1

  • -----
  • Posts: 632
  • Joined: 28 Aug 2014
  • Loc: Kansas, USA

Posted 22 July 2021 - 12:32 PM

I'm continuing to experiment, but so far it looks like the total integration time (time of all the subs added up) is what's really important. I used to use 8 minute subs, then shot for years at 6 minute subs, then went down to 5 minute subs.  Now I have standardized on 4 minute subs, and have found that 1 1/2 hours of 4 minute subs gives FAR better results than 1 hour of 6 minute subs.

Next, I'm going to try 1-1/2 to 2  hours worth of 3 minute subs and see what results I get. Also, previous point well made: it's less frustrating to lose a 3 minute sub to a satellite trail than to lose a 10 minute sub!

Ron Abbott


  • bobzeq25 likes this

#17 zoltrix

zoltrix

    Vostok 1

  • -----
  • topic starter
  • Posts: 102
  • Joined: 23 Feb 2021

Posted 22 July 2021 - 01:08 PM

A good mount remains important.  Autoguiding is just about essential, unless you have a really expensive mount.  >$5000.  And maybe still a good idea then.
 

.


This is not clear to me
A good mount and autoguiding can keep the target in the center of the camera but this should not be strictly necessary with short exposures since the frames can be aligned via software
Of course there is a certain tollerance range but this can make the difference between the need of a top mount and the possible use of a normal one

#18 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 25,653
  • Joined: 27 Oct 2014

Posted 22 July 2021 - 01:52 PM

This is not clear to me
A good mount and autoguiding can keep the target in the center of the camera but this should not be strictly necessary with short exposures since the frames can be aligned via software
Of course there is a certain tollerance range but this can make the difference between the need of a top mount and the possible use of a normal one

Even with short exposures your eccentricity and full with at half maximum are certainly better with autoguiding.  Which is a big factor.  Why the vast majority of experienced imagers autoguide

 

A high quality mount is a smaller factor.  But still relevant if you have a good scope.  People don't spend the money on good mounts because they're a status symbol, like a Rolex.



#19 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,984
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 22 July 2021 - 09:49 PM

Hello

Probably this question has been asked several times
Is it true that,a long,let's say,10 minutes exposure is equivalent to 20, 30 seconds short exposures, thanks to the modern CMOS cameras with low read noise ?
Are significant comparative tests, all things being equal, available ?

So we need to be clear here.

 

As long as you get TEN MINUTES TOTAL with a CMOS camera, AND so long as you are swamping the read noise by the same relative factor (i.e. signal is 10x larger than read noise squared)...

Then, yes, 30 second subs will produce end results that are just as good.

 

Make sure you read that carefully...because, in no way, no matter how you slice it, is a SINGLE 30-second CMOS sub as good as a SINGLE 10-minute CCD sub. I just want to make that imperatively clear here, since there has been so much confusion around this topic over the years. When there are differences in total signal, barring HUGE differences in noise, the larger signal wins. Peroid. Ten minutes gets you a lot more signal than 30 seconds. Period. 

 

Total exposure and noise swamping. If the total exposure times are equivalent, and you are swamping the read noise by at least the same factor, then a low noise camera will produce results at least as good as CCD (barring other quality factors). Now, keep in mind, another thing you HAVE to do, in order to properly COMPARE such exemplars, is to NORMALIZE the results. That means downsamlping the higher resolution image to the same effective pixel size as the lower resolution image. Basically, take the longest edge, and re-scale the higher res image to have the same number of pixels along the longest edge. Without normalization, then you cannot perform a proper comparison.


Edited by Jon Rista, 22 July 2021 - 09:53 PM.

  • bobzeq25 and Dynan like this

#20 iantaylor2uk

iantaylor2uk

    Explorer 1

  • -----
  • Posts: 93
  • Joined: 19 Dec 2020
  • Loc: Chester, UK

Posted 23 July 2021 - 04:05 AM

It is pretty obvious that a 30 second sub won't be as good as a 10 minute sub, but what about 20 30 seconds subs that are stacked? Surely the result should be similar to that of a single 10 minute sub.

#21 Morgar

Morgar

    Explorer 1

  • -----
  • Posts: 54
  • Joined: 22 Jun 2021

Posted 23 July 2021 - 04:39 AM

No, there is no linear equivalence, there is an efficiency curve that increases rapidly and decays with time, in the number of subs the "peak" would be between 20 and 40. In the exposure time it depends on the material used, the object to photograph, and the environment. Between 2 and 10 minutes is a reasonable time. So for equal integration time, it is as inefficient 100 x 30", than 3 x 1000". You will always see that everyone uses a balanced integration, something like 10 x 300" or 25 x 120".

 

That said, I have had a modern CMOS for some time but from the tests I have done in general increasing the integration time is more effective than increasing the number of subs.



#22 Tapio

Tapio

    Aurora

  • -----
  • Posts: 4,702
  • Joined: 24 Sep 2006
  • Loc: Tampere, Finland

Posted 23 July 2021 - 05:30 AM

I'd like to add that dithering (in most cases) require autoguiding.
Without autoguiding just drifting is not as good as auto guide-dithering.

#23 iantaylor2uk

iantaylor2uk

    Explorer 1

  • -----
  • Posts: 93
  • Joined: 19 Dec 2020
  • Loc: Chester, UK

Posted 23 July 2021 - 05:30 AM

Well it depends on focal length and whether you are imaging narrowband or broadband. There are plenty of imagers out there with fast systems (f/4 or lower) who are getting good result by taking hundreds of images at exposure lengths of 30 seconds or lower. You also have to take account of the effect of dropping substandard frames. It is quite a hit if you have to drop a few 10 minute subs (due to satellites or other reasons) whereas dropping a few 30 second subs is not a big deal.
  • Dynan likes this

#24 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 25,653
  • Joined: 27 Oct 2014

Posted 23 July 2021 - 08:51 AM

It is pretty obvious that a 30 second sub won't be as good as a 10 minute sub, but what about 20 30 seconds subs that are stacked? Surely the result should be similar to that of a single 10 minute sub.

That's the question involved in deciding subexposure.  How can I get the best possible image out of 10 minutes (more of course) of total imaging time.  The answer is rarely one sub (it can be, narrowband).  Broadband it's almost always some number of subs. 

 

The answer fundamentally depends on read noise, light pollution level and F number.  This table illustrates, for one camera.  It goes all the way from <10 seconds, more than 360 subs, to  10 minutes (actually 15), with skies of 21.6 mag per arc sec, gain zero, and F10.  Varies by a factor of more than 60.  Shows clearly why copying someone else's subexposure is a bad idea.

 

https://www.cloudyni...xposure tables



#25 Poynting

Poynting

    Mariner 2

  • *****
  • Posts: 294
  • Joined: 01 Apr 2012
  • Loc: San Antonio, TX

Posted 23 July 2021 - 09:31 AM

I appreciate the healthy discussion so far. I've been thinking about this lately as I have standardized at 5min subs at FL around 1000mm, f/7, CEM60 guided, but my keep rate is 50-60%. I'd like a better keep rate, but fear going down in exposure time I will lose out on finer detail in my image. I guess I will have to run a test to be sure.




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