Jump to content


Photo

5 or 30 minute exposures for narrowband?

  • Please log in to reply
239 replies to this topic

#1 cn register 5

cn register 5

    Viking 1

  • -----
  • Posts: 760
  • Joined: 26 Dec 2012

Posted 27 January 2013 - 06:47 AM

What the benefits (if any) of doing 15 to 30 minute exposures for narrowband compared to shorter exposures, such as 5 minutes?

The only advantage I can see is that fewer exposures will have less read noise but that's it. The other sources of noise are proportional to the total exposure time aren't they?. Read noise is pretty small compared to everything else isn't it?

The disadvantage I can see are that if something goes wrong more is lost with long exposures.

Chris

#2 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 27 January 2013 - 07:33 AM

It's all about signal. Narrowband blocks a lot of the light spectrum and requires more time to accumulate enough signal. Also the number of frames you can stack to improve your SNR is limited before benefits disappear. With any image, the longer the better to a point. If you have a lot of sky glow, that is signal as well and well overwhelm your background. One reason guys can shoot narrowband in light polluted areas is they can block a lot of that sky glow. You could take 30 5 min NB images and 5 15 min shots, and the 5 15 min shots will look much better than the 30 5 min shots even though you have less total time. You want your exposure to go until you've reached your sky noise limit. In my case I'm at a dark site, so my typical NB image is 30-40 for each. For an expert explanation read John Smith's article here: http://www.hiddenlof...ubExposures.pdf

#3 dawziecat

dawziecat

    Vanguard

  • *****
  • Posts: 2226
  • Joined: 20 Oct 2010
  • Loc: Rural Nova Scotia

Posted 27 January 2013 - 08:23 AM

There is theory . . . and there is what YOU can do in YOUR world with YOUR equipment under YOUR skies!

I quite accidentally discovered that I could get some nice results with (hold your breath) ONE MINUTE NB exposures!

It amazed me!

I've posted the attached image here before. It was taken with a 180mm lens at f3.5 and a SBIG ST-8300M camera with a 3nm Ha filter.

I'm not saying it's ideal to do this. Not hardly! But, if short exposures are the best your gear can do, you can still get some pretty nice images.

Attached Files



#4 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 27 January 2013 - 08:57 AM

Using a very bright DSO is hardly a comparison. There's so much signal there already, you don't need long exposures. If your're content with only shooting mag 4 or greater images and your optics are very fast, you can shoot short exposures. The key point is signal. The problem is you're going to run out of targets very quickly.

#5 Inverted

Inverted

    Mariner 2

  • *****
  • Posts: 210
  • Joined: 19 Jan 2013
  • Loc: LP Land

Posted 27 January 2013 - 09:26 AM

My understanding is it depends on the read noise of your camera and also some thing like tracking. Most noise sources, sky noise, dark noise etc.. really just depend on exposure time. But it doesn't really matter if it is one long exposure, or lots of little exposures, the noise is just dependent on the sum of the exposures. However, every time you take an image, you get some read noise from the camera. This does not increase with time though, it increases with the number of exposures. If you have a camera with really low read noise, then it shouldn't matter as much if you take one exposure or lots of exposures. From the specs, some of the new cameras are getting really low. The 8300 has about half of the read noise, that we considered "low" just a few years ago, and some of the Atik cams, for example, have half that!

So now, when you take a long exposure, your also more prone to image shift, tracking error etc, that blur the image and kill some detail. The purpose of a longer exposure though, is to bring out that detail, by increasing SNR.

Not all setups can handle really long exposures as well as well as others though. I think for most setups, there is a happy medium between what the equipment can handle and the minimization a of camera reads and therefore "read noise". Not everyone has a Paramount mount, and zero flex in their optical train etc... And not everyone will see a difference between a super long exposure and lots of shorter exposures. Think people are best advised to experiment and see what works best for their particular equipment combination.

#6 freestar8n

freestar8n

    Soyuz

  • *****
  • Posts: 3890
  • Joined: 12 Oct 2007

Posted 27 January 2013 - 09:56 AM

Everything depends on the read noise of the camera and the amount of sky glow given your site, the light pollution/moon, and how narrow your filters are. You can leave "signal" out of the discussion and just focus on the noise terms - and the goal is to have the sky noise be much greater than the read noise. If you have dark skies and high read noise you will find more benefit from longer subs. If you have bright skies and low read noise, you won't.

f/ratio plays a role because a faster system will accumulate sky photons faster per pixel than a slow one. Again - this is independent of the nebulosity signal itself and only depends on the "black" parts of the sky - but it means faster systems can use shorter subs.

If your guiding and polar alignment aren't ideal then that would bias things toward the shorter exposures - due to guiding and resolution rather than noise.

It's good to know the actual read noise of the camera and how much signal is coming from sky glow rather than nebulosity. A few exposures should give a feel for these values.

Frank

#7 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 27 January 2013 - 10:19 AM

Well Frank, for once I agree with you. You can find a calculator here; http://www.stanmoore...ack_photom.html , as well as discussions. I hate it when people mislead by saying time is time 10 1 min exposures does not equal 1 10 min exposure. There's also a limit to how many exposures you can take before there is no increase in SNR and you're wasting your time. That's why Hallas says "take as long of exposures as your equipment will allow" You don't have to have a Paramount to take long exposures, it just makes it easier. JWalk is prime example of someone who could take a CGEM and take wonderful images. Why? Because he had it hypertuned then learned how to properly polar align and guide.

#8 Inverted

Inverted

    Mariner 2

  • *****
  • Posts: 210
  • Joined: 19 Jan 2013
  • Loc: LP Land

Posted 27 January 2013 - 10:35 AM

Just another though, as I sit here, playing with some test images, taken last night to try out my new setup.... I'm looking at some 4 min, versus 10 minute m81 pics,. In the 4 minute exposures, I can use the histogram to bring the core to a pinpoint. However, with the 10 minute, the core is saturated and at best maybe 20-30x the size. Granted the setup did handle the 10 minute exposure fine, the stars don't show more trailing in the 10 minute, there is better detail in the arms etc... But the minimum resolution of bright objects, just isn't there.. Of course, you could always take both long and short exposures and blend them to get certain detail, which may end up more efficient than taking lots of short exposures, but as mentioned, it depends,

#9 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 27 January 2013 - 11:16 AM

Just another though, as I sit here, playing with some test images, taken last night to try out my new setup.... I'm looking at some 4 min, versus 10 minute m81 pics,. In the 4 minute exposures, I can use the histogram to bring the core to a pinpoint. However, with the 10 minute, the core is saturated and at best maybe 20-30x the size. Granted the setup did handle the 10 minute exposure fine, the stars don't show more trailing in the 10 minute, there is better detail in the arms etc... But the minimum resolution of bright objects, just isn't there.. Of course, you could always take both long and short exposures and blend them to get certain detail, which may end up more efficient than taking lots of short exposures, but as mentioned, it depends,


On a very bright DSO such as M 81, you are going over saturate the core with long exposures. Short exposures will leave out all the detail in the rest. The best method is to combine short and long. In my example below, M 31 was shot at 15 min exposures and 2 min exposures. The core was blended into the image in using a reveal all mask and feathering as a layer.

Attached Files



#10 Inverted

Inverted

    Mariner 2

  • *****
  • Posts: 210
  • Joined: 19 Jan 2013
  • Loc: LP Land

Posted 27 January 2013 - 02:45 PM

Buitiful image! Also, by the way, this is I assume this was probably infered, but to clarify. As we are talking about narrow band, the range will tend to be lower, so the odds of saturating anything is lower, so, you'd be less likley to benefit from shorter exposures for NB. So, that would be more justification to go longer.

#11 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 27 January 2013 - 05:11 PM

For narrowband imaging, in most cases read noise IS the dominant source of noise. With sufficiently narrow filters the read noise exceeds the shot noise from the background skyglow.

For me, I want to balance the benefits of shooting longer exposures (injecting a smaller multiple of the read noise into the final stack for a fixed total integration time) against the desire to have enough subs to capitalize on dithering by using an outlier-rejection method for image integration, removing uncalibrated hot pixels, cosmic ray strikes, satellite trails, etc. I have therefore standardized on 20 minute narrowband exposures and 3 hours minimum per channel, which means I'll have at minimum 9 frames to stack.

Longer would be better. If I planned 5 hours of exposure, I'd probably do 30 minute subs because then I'd have 10 to stack.

All this assumes that your mount and guiding solution is capable of exposing for arbitrary lengths of time without trailing. If that's not the case, your hardware will set an upper limit on your exposure time.

#12 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 27 January 2013 - 05:22 PM

There is theory . . . and there is what YOU can do in YOUR world with YOUR equipment under YOUR skies!

I quite accidentally discovered that I could get some nice results with (hold your breath) ONE MINUTE NB exposures!


Do you happen to have links to full-resolution versions of these shots, ideally without noise reduction? It's really hard to judge whether the 1 minute exposure versions are noisier at the image scale you posted.

#13 cn register 5

cn register 5

    Viking 1

  • -----
  • Posts: 760
  • Joined: 26 Dec 2012

Posted 27 January 2013 - 05:56 PM

Thanks to everybody for the information, especially the paper by John Smith and the images.

What I'm hearing is that the idea is to make the read noise a small part of the contribution to each image and that the time this takes depends on the amount of other noise, such as sky glow.

In my case I'm in the UK, on the edge of a town, so am not getting the sort of dark skies you get in New Mexico Skies - in fact it's often more cloud glow than sky glow.

I haven't measured it but would guess that I can get away with more like 5 minutes for now, even with narrowband filters. In any case I'd rather stack 10 out of 12 5 minute exposures than loose two out of three 20 minute exposures because of clouds or passing planes.

Thanks for your help,

Chris

#14 dawziecat

dawziecat

    Vanguard

  • *****
  • Posts: 2226
  • Joined: 20 Oct 2010
  • Loc: Rural Nova Scotia

Posted 28 January 2013 - 12:08 AM


Do you happen to have links to full-resolution versions of these shots, ideally without noise reduction? It's really hard to judge whether the 1 minute exposure versions are noisier at the image scale you posted.


Original posting.

Rest assured, theory IS correct! :) The longer exposures DO produce a smoother image than the ones fifteen times shorter! :bigshock:

Point is, people who don't have high $$$ mounts can still get some nice pics with rather shockingly short exposures . . . even with NB filters.

#15 wolfman_4_ever

wolfman_4_ever

    Apollo

  • -----
  • Posts: 1245
  • Joined: 15 Jul 2011
  • Loc: El Segundo, Ca, So. Cal

Posted 28 January 2013 - 12:56 AM

all great posts here!

#16 J.P.M

J.P.M

    Gemini

  • *****
  • Posts: 3102
  • Joined: 14 Nov 2006
  • Loc: 65"01'N25"47'E

Posted 28 January 2013 - 06:24 AM

Just my 2 euro cents...

Longer is better since, if there is missing some very weak signal in short exposure, it doesn't show, up no matter how many of them you going to stack. Specially in NB imaging. Usually I'm using 1200 seconds subs but now and then, I have gone up to an hour, if the target is an extremely dim one.

#17 freestar8n

freestar8n

    Soyuz

  • *****
  • Posts: 3890
  • Joined: 12 Oct 2007

Posted 28 January 2013 - 07:38 AM

That would work if the limiting noise source is read noise, but if sky glow dominates the read noise then a sum of 2 5-minute frames won't look much different from a single 10-minute frame. Within the assumptions of typical models for ccd imaging SNR, read noise is the only thing you can reduce by going longer - and if it's a small part of the overall noise, then the only way to pull out faint signal is with longer total integration time - and longer subs won't help much.

The ccd noise models could be wrong or incomplete - but in that case it would be nice to know what is missing - so it can be added to the model.

Frank

#18 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 28 January 2013 - 09:18 PM

I had a discussion today with a narrowband expert. With Narrowband you are not likely to ever exceed the sky glow level so taking the longest images possible is very important. The longer exposures bring out much more nebulosity that will never be seen with short exposures. There is no measurable improvement in image quality once your stack goes beyond 16, therefore longer images become even more important.

#19 freestar8n

freestar8n

    Soyuz

  • *****
  • Posts: 3890
  • Joined: 12 Oct 2007

Posted 29 January 2013 - 02:25 AM

I would be interested in any numbers you can provide, along with an appropriate noise model, to help make your (his?) case. I consider it pretty obvious that a fast lens with a wide passband, a ccd with low read noise, and a full moon would likely be limited by Poisson noise from the sky background rather than read noise - and not benefit from long exposures. I welcome you to invite the person you are citing to explain his reasoning.

The importance of read noise and longer subs depends entirely on the scale of read noise compared to other noise sources. Without those numbers, you can't say if longer subs would help much - and as described by several people above, they may make the end result worse due to imperfect guiding and having fewer subs for a good statistical combine with noise rejection.

Frank

#20 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 29 January 2013 - 04:24 AM

Frank I know you have a lot knowledge, but this gentleman happens to be an optics expert. I will quote from his website;

Also, narrowband signals are inherently weak, requiring long exposures of 20 – 40 minutes.


#21 dawziecat

dawziecat

    Vanguard

  • *****
  • Posts: 2226
  • Joined: 20 Oct 2010
  • Loc: Rural Nova Scotia

Posted 29 January 2013 - 08:45 AM

Also, narrowband signals are inherently weak, requiring long exposures of 20 – 40 minutes.


Yes, but at what focal ratio?

You most certainly do not require such exposure durations with f/2.8 optics. Lots of iconic objects in astronomy are not small in angular diameter and not especially faint in Ha either.

These can be imaged very satisfactorily in NB with fast camera lenses and almost all of us do image these objects when starting out in this hobby.

Sure, it is ideal that one can do long subs and great if your gear is up to it. But, for a great many of us, our gear limits what we can reasonably accomplish.

My biggest thrill in this hobby to date was when I first tried an Ha filter with a DSLR and a 300mm f/2.8 camera lens. No one should deny themselves this out of fear their gear is not up to 20-40 minute exposures. Stack up a dozen or two exposures of a few minutes each (or even less) and be prepared for a kick!

Seems to me there is a conflict of expectations here. What you need to image a small SNR in NB and what you need to do The Rosette, The NA Nebula complex, the Horsehead, and the like are pretty different in regards to FL, associated focal ratio and attendant sub exposure length.

#22 vpcirc

vpcirc

    Soyuz

  • -----
  • Posts: 3963
  • Joined: 09 Dec 2009
  • Loc: Merced CA

Posted 29 January 2013 - 09:07 AM

That's a great example, he said to try and image the horsehead at short exposures and see what happens compared to longer exposures. A very bright DSO like the Rosette isn't going to show the signal problem because so much is there.
But again the point is why stack 2 dozen images. There is zero gain from going beyond 16.

#23 David Pavlich

David Pavlich

    Transmographied

  • *****
  • Administrators
  • Posts: 27206
  • Joined: 18 May 2005
  • Loc: Mandeville, LA USA

Posted 29 January 2013 - 09:10 AM

Remember to play nice and leave the personal comments out of the discussion. It adds nothing to the conversation.

David

#24 Inverted

Inverted

    Mariner 2

  • *****
  • Posts: 210
  • Joined: 19 Jan 2013
  • Loc: LP Land

Posted 29 January 2013 - 09:16 AM

I'm no expert on CCDs or ccd noise models. However, just to state the obvious, it seems clear that if read noise is the dominant source of noise, and read noise varies considerably between cameras, then it would seem that the same snr, over a fixed integration time will vary considerably by camera. For example, if you have a camera with 3e- vs one with 15e- , take 1 hour of integration, with 6x10 minute exposures, with the 15e- camera, then the same read noise contribution, with the 3e- camera should be archived with 30x2 minute exposures. However, through sampling, all random noise should tail further towards the asymptote with the 30x2 minute exposures. Also, if you either your shots, fixed pattern noise should as well. So, your total snr I think may actually be better, as long as the total read noise contribution is roughly the same or better. also, depending on how low the read noise is and how close it is to the point where read noise is no longer the dominant noise is, it should also start to become more beneficial to "sample out" other sources of noise, by taking more exposures. I have no idea what that level is though, or how close modern cameras are getting to it.

Anyways though, still in practice, you may well be better off taking longer exposures wot the low e- cameras, but as mentioned, I'm sure there a lot of factors to consider, even without considering camera noise models.

#25 Inverted

Inverted

    Mariner 2

  • *****
  • Posts: 210
  • Joined: 19 Jan 2013
  • Loc: LP Land

Posted 29 January 2013 - 09:30 AM

There is zero gain from going beyond 16.


No doubt gains decrease as you go, but anyone who understand statistics and the central limit theory, should understand that there is alway a gain until you reach the asymptote. Again, i'm no expert on ccds, but do know enough to know there will be a gain... Whether the gain improves snr noticeabley depends on all sorts of factors, but I would be interested to hear how he came up with 16. I suspect that is a rough example, based on a certain setup, sky conditions, target brightness etc.. That may be true for that one instance,, but clearly is not generalizeable. I know I can certainly see a difference between a statck of 16x5 min exposures and a stack of 50... Whether or not the stack of 50 would be better than a single 250 minute exposure, I don't know "generally" but with my setup the 50x5 would be better LOL






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics