Different exposure times vs same exposure times

[tingting44]

Wow, The things one learns.

Thankyou Olly for pointing us to the above code. I have installed a trial version in my CS3 extended and without any effort I produced :

M31.jpg

After a bit more testing it looks as if I'll be purchasing.

Jeremy

don't forget to crop out stacking artefact's from the edge of your final stack, it messes up your histogram when adjusting levels, yes GXT is excellent, I would have purchased it all ready if I did not plan on getting PixInsight as soon as I have some spare astro funds.....

[JRWASTRO]

don't forget to crop out stacking artefact's from the edge of your final stack, it messes up your histogram when adjusting levels, yes GXT is excellent, I would have purchased it all ready if I did not plan on getting PixInsight as soon as I have some spare astro funds.....

Greetings Martin,

This is not my image but one previously posted on this thread.  I just used the image as a quick test and have just paid for the licence as this is " my type of software" ... no learning curve!  Yes I get LP but it is not bright pink but a terrible shade of yellow!  Had a look at PIXINSIGHT but balked at paying "that much". When my imaging and  processing capabilities over take the current software I'm using I might consider spending the cash. And thankyou for the tips... all is welcome.

Kind Regards,

Jeremy.

[ollypenrice]

I know PI isn't given away but, to be fair, it represents phenomenal value for money and includes free updates. By comparison many other astro programmes are exhorbitant. AstroArt is another pleasant exception and is top value.

Olly

[lensman57]

I know PI isn't given away but, to be fair, it represents phenomenal value for money and includes free updates. By comparison many other astro programmes are exhorbitant. AstroArt is another pleasant exception and is top value.

Olly

I second this. I too was unsure about the value of this software as it is not cheap but now I am convinced having had it now for nearly a year. As Olly said once the DBE is worth the asking price alone. I also like StarTools for its power and simplicity V price but this is an image developing software only as PI does lot more such as aligning, integration and on. 

A.G

[vertigo262]

So tonight I went out, to a better location which is much darker. I wanted to get Orion Nebula but, for some reason this is all I got. which shows nothing. which confuses me because I've never had a problem getting anything even with 30 second exposures. I don't think it was my alignment because I took some pictures of horse head, andromeda, heart showing some faint colors and they were in frame on slew.

But for some reason I could not get orion nebula to show up. My only thought is that it wasn't visible tonight. But I don't know if that is possible. I could see Orion constellation as clear as can be.

I also did some tests on andromeda, this time 200 seconds. But my polar alignment was only good for about 120 seconds. And I need to figure out what I am doing wrong with alignmaster.

Anyways, here are the pics. I'm getting a little better. But if anyone can explain why orion didn't show up, that would be helpful

[tingting44]

good one on getting out and getting some data, as for orion it did not show up i would say is because you are not on target, if you were on it, orion would show with a few seconds exposure

[Rico]

Sorry to piggyback on this thread, but going on from the discussion about the different exposure times ... How much better is a long exposure vs a stack of shorter exposures which collectively have the same duration as the longer one?

E.g 10x600s vs 60x100s?

[D4N]

Vertigo you just missed the Orion Nebula, it's off to the right a bit.

You can use astrometry.net to plate solve images pretty quickly so I did yours.

The M31 has something else going on not just poor PA as there is a dog leg in the star trails.

Did you walk past the mount during capture?

If you take longer exposures the idea is you improve the signal to noise ratio.

It is also nice to see a result from a single sub before processing

However you will also have more data ruined by sats as if there is one sat in 20 mins you would lose one out of two ten min exposures but just one out of 40 30 second exposures.

Sent from my iPad using Tapatalk

[ollypenrice]

Sorry to piggyback on this thread, but going on from the discussion about the different exposure times ... How much better is a long exposure vs a stack of shorter exposures which collectively have the same duration as the longer one?

E.g 10x600s vs 60x100s?

Oh boy, can of worms. I'm not a theorist. I take pictures - lots of them, be it said. My view is that if you have a dark site (and a cooled camera) you should take long subs, then longer, then longer still. I feel that long subs blow short ones clean out of the water. I confess that I don't follow in detail the theoretical arguments for the equivalence of lots of short subs because my night in, night out experience says exactly the reverse. I am totally certain that long subs demolish short ones. Those who collaborate with me here, notably Yves Van den Broek and Tom O'Donoghue (both APOD imagers) take the same view. Indeed it was Yves who first said, 'Why don't we go for 30 minute luminance subs?' and we did. And I can say this for sure; we won't be going back!

I have never imaged in light pollution. I've processed lots of images captured in LP but I'll leave it to others who have experience in LP imaging to talk about that.

Olly

http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/22435624_WLMPTM#!i=2266922474&k=Sc3kgzc

[RichM63]

Dou you understand the business of stretching in Levels and Curves? This is a great explanation by RobH on here:

http://www.middlehillobservatory.co.uk/articles-primers/Levels%20and%20curves.htm

Olly

Hi Olly does this guy have a website, this tutorial is so easy to understand, I'm hoping he has more.

Cheers,

Richard

[ollypenrice]

Hi Olly does this guy have a website, this tutorial is so easy to understand, I'm hoping he has more.

Cheers,

Richard

He does indeed and here it is. http://www.middlehillobservatory.co.uk/

He's also on here fairly regularly.

Olly

[RichM63]

Great, thanks, clear and concise, answers or should I say explains various aspects.

I'm no AP imager 'yet', but even I can see that 10x500 seconds is more data than 500x10 seconds.

Shorts bursts of 10 seconds on the same target will only acquire the same data over many iterations and would seem to also multiply the mechanical problems within the equipment etc.

Whereas fewer longer hits will absorb more data and should reduce mechanical issues within the train.

ie 500 seconds worth every time, rather than 10 seconds worth every time.

The 'secret' appears to be how we use the captured data.

Hopefully I have understood the concept.

[ollypenrice]

Long subs do bring their own challenges, notably guiding; you need to have this taped. Things like satelite trails and planes are not such a problem. Sigma reject will often 'disappear' them and, if not, you do a stack with trails and a shallower stack without. Put the shallower stack as a lower layer in Ps and the deeper, trailed stack on top. Erase just the trails from the layer and flatten the images. Best of both worlds apart from just under the trails, and will this show? Doubt it.

The thing about long subs is that they get the faint signal above the read noise. But if I go down this road I'll get all sorts of proof that this isn't so and I'm really not interested since I know I'll be sticking with long subs!

Olly

[mrjonathanbrown]

Hi guys I have been following this thread and I just have a quick question, I never thought to much about this until I read the thread but I have been reading some AP images are exposed between 30 - 90 minutes is that one single shot or is that shots stacked together adding up to 30/90mins? (Hope that makes sense) 

[ollypenrice]

It could be either, though only a few rather expert imagers go to 90 minute sub exposures. A sub exposure (or 'sub') is a one of a number of similar exposures stacked together to produce a final image. If yu stack a number of sub exposures of whatever length then the random noise in any one sub is averaged out for a cleaner final result.

Olly

[mrjonathanbrown]

Yeah I was finding it difficult understanding it at first because I thought 90min exposure would be over exposing causing the image to burn but on SGL you are always learning. 

Is it really a matter of trial and error talking from a beginners point of view, Messing about with different exposure times..

Olly, I snuck on your page to view some images I'm really annoyed! They are too good. 

[Rico]

Oh boy, can of worms. I'm not a theorist. I take pictures - lots of them, be it said. My view is that if you have a dark site (and a cooled camera) you should take long subs, then longer, then longer still. I feel that long subs blow short ones clean out of the water. I confess that I don't follow in detail the theoretical arguments for the equivalence of lots of short subs because my night in, night out experience says exactly the reverse. I am totally certain that long subs demolish short ones. Those who collaborate with me here, notably Yves Van den Broek and Tom O'Donoghue (both APOD imagers) take the same view. Indeed it was Yves who first said, 'Why don't we go for 30 minute luminance subs?' and we did. And I can say this for sure; we won't be going back!

I have never imaged in light pollution. I've processed lots of images captured in LP but I'll leave it to others who have experience in LP imaging to talk about that.

Olly

http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/22435624_WLMPTM#!i=2266922474&k=Sc3kgzc

Thanks Olly. I managed to convince myself that averaging short subs improves the SNR, and as a result allows for better stretching. However, i think the equivalence to longer exposures is probably not in averaging the stacks but instead the 'summation' of these stacks. Unfortunately while more luminance is being added, so is the noise and so there is absolutely no improvement to the SNR. My idea of summing, stems from the idea of collecting photons, however, realistically, adding images doesn't work since the image will simply become completely saturated.

[ollypenrice]

I think it's remarkably hard to saturate images with a CCD camera. Star cores over expose, yes, and particularly in luminance, but you don't have to apply the luminance to the stars.

Olly

[sharkmelley]

Oh boy, can of worms.

Indeed   But I'll risk jumping in anyway!

I'm no AP imager 'yet', but even I can see that 10x500 seconds is more data than 500x10 seconds.

No.  In both cases you have captured 5000 seconds worth of photons.  It's the same amount of data.  The difference is that in the first case you have captured 10 lots of read noise vs 500 lots of read noise in the second case.  So 10x500 secs should result in a less noisy image than 500x10. 

Unfortunately while more luminance is being added, so is the noise and so there is absolutely no improvement to the SNR.

That's not right.  The luminance grows faster than the noise does (because the noise is random), so the SNR improves.  This link explains how exposures are combined to reduce SNR: http://starizona.com/acb/ccd/advtheoryexp.aspx

I confess that I don't follow in detail the theoretical arguments for the equivalence of lots of short subs because my night in, night out experience says exactly the reverse. I am totally certain that long subs demolish short ones. Those who collaborate with me here, notably Yves Van den Broek and Tom O'Donoghue (both APOD imagers) take the same view. Indeed it was Yves who first said, 'Why don't we go for 30 minute luminance subs?' and we did. And I can say this for sure; we won't be going back!

I am surprised that you notice an improvement with 30min subs in luminance.  The idea behind long exposures is so the backgound noise from the skyglow drowns out the read noise (e.g. by a factor of 20).  That is the principle behind subexposure calculators such as this one: http://www.ccdware.com/resources/subexposure.cfm   Hence for Narrowband, long exposures are important because the recorded skyglow is so low that it takes a long time to record enough skyglow to drown the read noise.  However, for luminance, the recorded skyglow would be very much higher, so the read noise is drowned out much quicker.  Once your exposures are long enough to drown out the read noise then the difference in various exposure lengths will not be noticeable in the stacked image (for the same total exposure time in the stack).

From an experimental point of view, it would be interesting to take the background level from the 30min luminance sub and put it through the exposure calculator to see what exposure length it suggests. 

It is always possible there is some other additional effect taking place.

To be clear, longer exposures will always be better.  It's just a question of how noticeable it is in the final stack (given the same total exposure time).

Mark

[ollypenrice]

This is why I don't get involved in theoretical discussions! I don't know why 30 minute subs beat 15 minute ones but I just find that they do. Tim, who is very expert on faint planetaries, has tried subs of several hours. I do work from a dark site, though, SQM 21.9 on a good night.

Using 15 minute L subs I found the outer halo of M31 shapeless, yet I'd read about the fact that it has a 'turned up' end which has a name - though I can no longer find this reference. Sorry about that!  However, 7x30 minutes at F5 and 3.5 arcsecs per pixel found it very easily, viz;

http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/i-xbvjFDF/0/X3/M31%20Outer%20HaloLHE-X3.jpg

Also Tom and I tried to find colour in the IFN and did all our LRGB in 30 min subs.

We think it's brown!  But who knows??

Olly

[dph1nm]

The thing about long subs is that they get the faint signal above the read noise. But if I go down this road I'll get all sorts of proof that this isn't so and I'm really not interested since I know I'll be sticking with long subs!

Actually I am not surprised. I did a quick calculation as I happen to know that from an astronomically dark site 45 sec g-band exposures with 0.25" pixels on a 1.8m telescope are significantly affected by 5e- read noise. A 0.2m (~8") scope will need about 80 times the exposure to collect the same number of photons. So 1hr exposures with such a scope would still be read noise limited. Larger pixels will reduce this of course, but  I think you are still talking substantial sub lengths before you can get over the effects of read-noise with amateur-sized scopes at dark sites.

For those of us with skies 5 magnitudes brighter this isn't such an issue of course!

NigelM

[ollypenrice]

Actually I am not surprised. I did a quick calculation as I happen to know that from an astronomically dark site 45 sec g-band exposures with 0.25" pixels on a 1.8m telescope are significantly affected by 5e- read noise. A 0.2m (~8") scope will need about 80 times the exposure to collect the same number of photons. So 1hr exposures with such a scope would still be read noise limited. Larger pixels will reduce this of course, but  I think you are still talking substantial sub lengths before you can get over the effects of read-noise with amateur-sized scopes at dark sites.

For those of us with skies 5 magnitudes brighter this isn't such an issue of course!

NigelM

I'll be glad to confirm this if you'd care to lend me the 1.8 metre telescope in question...

lly

[sharkmelley]

Very good pictures Olly.  I've never seen quite that much halo around M31 and the IFN image looks postively bright.

I've just done some quick calculations and I've made the shock discovery that I ought to be using longer exposures than I presently do (using a DSLR on the Tak Epsilon at F2.8 )

I analysed the background of a 5 min ISO 800 sub taken at the recent Kelling Heath Star Party where I registered an SQM reading of 21.2 (it wasn't one of their darkest nights).  The number of electrons in the R,G,B channels are  380, 400, 180 respectively.  Taking the square root of those figures gives a background noise of approx  20, 20, 13 RMS electrons respectively. Given that the read noise of my DSLR is 5 electrons it means that the read noise is contributing a noticeable componet of noise to the final stack.  I should be using 15min exposures or more. 

Given that your sky is much darker, that you are using a slower scope (in terms of F-ratio) and you have low thermal noise on your camera, then 30 minute subs for L,R,G & B now seems perfectly reasonable to me.

Mark

[ianaiken]

I read somewhere an article, cannot remember who by now, that was talking about getting above read noise being the minimum exposure you want to aim for, and not necessarily the longest.  Is there a benefit under not so dark skies to go longer and use a high sub count overall.

[vertigo262]

ok, so I found it! I was a little off and just manually slewed the scope in desperation and found Orion Nebula. 

unfortunately I forgot my external laptop battery cable so I only got 16 subs which was really upsetting and I have to do it right with light, dark, flat, bias etc.

This is what I got, although noisy but I was somewhat happy for a beginner. Still figuring out DeepSkyStacker and photoshop.

16 x 60 seconds ISO 800

so I'm going out again to do it right. 

Question I have is, if I do multiple exposures, do I need to stack each exposure separate as different fit files, or can I just throw them all in the stack sequence?