Question about total exposure time

[Minimalist__]

I haven't done much imaging and am still very much a newbie, still getting to grips with the learning curve and terminology and techniques. I was hoping someone could clarify something for me.

My understanding of stacking is that it is used to 'average' out the data, thereby increasing the signal-to-noise ratio. The noise is random, but the signal is constant, therefore by stacking you keep all of the signal but remove the noise. First question is have I understood this correctly?

If I have understood this correctly then my second question is why do people (and also Deep Sky Stacker itself) say when given info about their images the total exposure time as the number of pictures x the exposure time e.g. 30x6mins 3 hours total exposure. If the stacking process produces an average of the 30 images in this example, then doesn't that mean the final image is just what a 6 minute exposure looks like when there's a high signal-to-noise ratio. Presumably a true 3 hour exposure of a subject would be radically different from 30x6min exposures stacked.

Thanks for clearing this up!

[Minimalist__]

Or to give another example of what I mean, since a 3 hour exposure seems a bit far-fetched, the resulting picture from 30x6mins exposure would (I think) look different from 12x15mins exposure, despite both being 3 hours total exposure time.

Furthermore does total exposure time even matter? When I look at the images from Gran Telescopio Canarias the exposure times are extremely short like 2x20secs because it has an extremely short focal length of f/1.6.

[Minimalist__]

Bah, f-ratio not focal length

[glappkaeft]

I'm going to try to answer the exposure time question first. The stacking question is related but I wasn't able to include it without complicating things even further.

Furthermore does total exposure time even matter? When I look at the images from Gran Telescopio Canarias the exposure times are extremely short like 2x20secs because it has an extremely short focal length of f/1.6.

Well without exposure time you won't get an image at all. This suggests that it 's important.    

I will omit all practical considerations (pixel saturation, sky glow, light pollution, etc.) and some minor theoretical ones (e.g. dark current) but the amount of image processing an image can sustain, how noisy it'll look and how faint details you can display depends on the Signal-to-Noise-Ratio (SNR). The SNR in astro images is equal to the square root of the Signal (the number of photons detected). There are three major factors that determine the amount of photons you'll gather:

a) Exposure time 

Quantum efficiency (QE), the fraction of the photons that reach the telescope that the camera actually detects (in addition to the camera this also depends on the quality/condition of the mirror/lenses used, any filters, etc.) 

c) Etandue or light gathering ability which is simply A*Omega = "Aparture area" * "Solid Angle per pixel" (the amount of the sky each pixel sees - basically pixel scale squared). This number can be seen as F-Ratio for digital photography. The F-ratio still of course works but only if we limit the comparison to pixels of the same physical size (Etandue is proportional to (pixel size/F-ratio)^2). 

The signal gathered is basically A*B*C. Improvements to QE ( are expensive and limited by the fact you cant do better than 100%. After a point improvements to Etandue tends to cost an insane amount of money (a decent 16" imaging rig costs tens of thousand of pounds and a Gran Teliscopico Canarias is 112 M£). So most amateurs tends to focus on exposure time instead. 

So why does everyone list exposure time? Measuring QE requires specialized equipment and very few people have heard of Etendue. Thus everyone lists the kit they used and the exposure times.

Hope this helps!

/Patrik

[Kirscovitch]

This is why I am sticking to observing and don't dare get into serious AP lol.

[Minimalist__]

[splice]

Hope this helps!

/Patrik

Thanks Patrik, it did help to get my head around things, and gave me new things to look into.

But my main query still remains - which is about the total exposure time. That is, is 30x6min exposures really an exposure time of 3 hours? Or is it just a total of 6 minutes exposure time with a high SNR due to the process of stacking?

In this way, the image produced from 30x6min exposures would be less easy to process and give less detail than the image produced from 12x15min exposures. People would list both these images as being 3 hours total exposure time, but I would expect the 12x15mins image to be higher quality (ie less noisy and processing to have produced less arterfacts and brought out more detail).

Thanks!

[ollypenrice]

We call three hours' exposure 'three hours' exposure' because... we exposed for three hours.   This seems eminently reasonable to me! You could call it, '15 minutes' exposure with the S/N ratio of  12x15 minutes exposure averaged out,' if you wanted to but don't expect it to catch on! The total exposure time and the length of subs tell other imagers something uselful and repeatabale.

Three hours with longer subs at a dark site, with a cooled camera, on a target that will not saturate the chip, will beat three hours ditto in shorter subs. Of this there is no doubt whatever. There are deniers but I can only insist that they are mistaken, and they are in a minority - not that that has any bearing on how right or wrong they are. One simple reason is read noise. You get more in 36 reads than you do in 15. 

If you want to quibble (strictly for forum fun!) about the term 'three hours' exposure' in the days of film, consider reciprocity failure. This would set in when the film emulsion was no longer able to give an increase in response proportional to an increase in time. If this set in after an hour but you had exposed for two hours, would this be a two hour exposure? Not in terms of signal.

If, like an idiot, I sit outside collecting 22 hours of data one thing is for certain sure; I 'm darned well going to call that 22 hours!!!    

Olly

[StuartJPP]

Yep, if you described your fantastic looking image and said that it was only a 10 minute exposure and not 30x10 minutes worth of exposure, arguments would ensue and someone would eventually get banned The important numbers are the individual sub lengths and the number of subs taken. With that knowledge and knowing the kit used it can the compared to like systems.

I also agree with Olly regarding the number and length of subs. Achieving a total exposure time that is the same i.e. made up from a different number and length of subs certainly does not equal the same end result. People keep harping on about it yet there has been no evidence to prove it, especially if you are talking about 1 second exposures vs. 10 minute exposures. There is just no way that the 1 second exposures can achieve anything like the 10 minute exposures for the same integration time. In fact I'd probably say that no matter how many 1 second exposures you have got, it will never beat a single 10 minute exposure (using the same camera!)...I wish to be proved wrong here people

[narrowbandpaul]

Like Olly and Stuart have said, it's the total integration time that quoted because that is how long you have exposed for. In simple terms, ignoring read noise, if you double the exposure time or use 2 subs frames then the SNR is the same. Again, it's the ideal world. This will never be realised due to read noise and dark current.

However the point is that for one, let's say you take 5min subs, then if you take 9 subs you have the same SNR as a 45 min sub. Clearly then you can relate the sub length and number of subs to the total integration time. Further the total integration time relates to the SNR .

So when folk say that they used 22 hours worth of 10min subs or whatever, it is equivalent (only in the ideal case that does not exist) to a SNR of one single 22hr sub.

The reality is that they are not equivalent but to work out the actual single sub integration time would be utterly tedious and quite pointless and tricky what with variable like read noise and dark current to account for.

So using multiple subs and increasing the exposure time are the same only for the special case of no dark current and read noise.

I point to this write up I did for the many short subs vs a few long ones debate. It may shed some more light.

http://stargazerslounge.com/topic/200397-many-short-exposures-or-a-few-long-ones/

Paul

[glappkaeft]

This is why I am sticking to observing and don't dare get into serious AP lol.

While the math and physics might help none of this stuff is required for AP. The common imaging rules of thumb and some experimenting also works perfectly good if one is not interested in the whys. Now if you want to do high quality absolute (calibrated) photometry...

[glappkaeft]

Sorry, yesterday I got roped into some last minute Xmas baking and cooking (we celebrate the 24th in Sweden). Santa left a new set of warm observing boots under the tree.

My understanding of stacking is that it is used to 'average' out the data, thereby increasing the signal-to-noise ratio. The noise is random, but the signal is constant, therefore by stacking you keep all of the signal but remove the noise. First question is have I understood this correctly?

It's close. It might be better to say that by stacking we add the signal and average out the noise, thereby increasing the SNR. Now in most cases (all but the simplest add) the stacking algorithm will rescale the image so that the pixel levels of the stack is similar to a single exposure. This averaging however does nothing with the data.

But my main query still remains - which is about the total exposure time. That is, is 30x6min exposures really an exposure time of 3 hours? Or is it just a total of 6 minutes exposure time with a high SNR due to the process of stacking?

In this way, the image produced from 30x6min exposures would be less easy to process and give less detail than the image produced from 12x15min exposures. People would list both these images as being 3 hours total exposure time, but I would expect the 12x15mins image to be higher quality (ie less noisy and processing to have produced less arterfacts and brought out more detail).

When it comes to noise the only difference between short and long exposures is caused by the extra number of read outs required by shorter exposures. Every time we read out an image a bias level is added to the data (to prevent the ADC from generating negative numbers) and this gets stacked along with the rest of the signals. Like all signals the bias signal has its own noise but usually (at least with cooled CCDs) it's possible to keep it at a negligible level by using a sufficiently long exposure time (when the photon signal is significantly higher than the bias level). As long this is achieved the stack of short images is virtually identical to one long exposure.

In practice shorter exposures have some advantages, e.g. fewer oversaturated pixels/stars and each ruined image (wind gust, tracking/guiding error, aircraft/satellite trails, etc.) costs you less exposure time. Picking a (or several different) exposure times is all about finding the compromise that works for your imaging target, purpose  and your equipment.

/Patrik

[lensman57]

I haven't done much imaging and am still very much a newbie, still getting to grips with the learning curve and terminology and techniques. I was hoping someone could clarify something for me.

My understanding of stacking is that it is used to 'average' out the data, thereby increasing the signal-to-noise ratio. The noise is random, but the signal is constant, therefore by stacking you keep all of the signal but remove the noise. First question is have I understood this correctly?

If I have understood this correctly then my second question is why do people (and also Deep Sky Stacker itself) say when given info about their images the total exposure time as the number of pictures x the exposure time e.g. 30x6mins 3 hours total exposure. If the stacking process produces an average of the 30 images in this example, then doesn't that mean the final image is just what a 6 minute exposure looks like when there's a high signal-to-noise ratio. Presumably a true 3 hour exposure of a subject would be radically different from 30x6min exposures stacked.

Thanks for clearing this up!

Hi,

The phrase stacking is hark back to the early 19th century when the pioneers of AP used to use glass plates exposed at long exposure times. They discovered that once these plates were stacked on top of each other  the very faint dark image of the stars ( silver halide) would add up and became darker and it was then usable  , the rest is history as they say. To put it in a simple way, your exposure time determines the strength of your singnal, stacking gets rid of the randomly distributed noise, therfore increasing signal to noise ratio as the noise is reduced. I do not believe either that 150 X 1 minute of exposures  is the same as 30 X  5 minutes of exposure. The large number of subs helps to reduce the noise in the stack to a geat extent but this will not cause an increase in signal as this is determined by the exposure time. There are other factors to consider but not relevant here. In the real world however, skies particularly in LPed urban areas is seeing limited and therefore there is a limit to how long an exposure will make any real increase to the signal level, particularly with DSLR and OSC CCDs, there is a point beyond which all that is added is sky fog which will actually decrease the signal. For my location and my equipment on an average imaging night this limit is about 600s at best, at times even 300 or 450s per sub have given good results. Olly can probably expose for 1800s per sub without any problems from his true dark site location for his LRGB imaging system, a great majority of us are not so privilaged.

Regards,

A.G

[ollypenrice]

In narrowwband, though, Tim images from Coventry in multi hour subs which is something I've never tried.

Olly

[russellhq]

There are other factors to consider but not relevant here. In the real world however, skies particularly in LPed urban areas is seeing limited and therefore there is a limit to how long an exposure will make any real increase to the signal level, particularly with DSLR and OSC CCDs, there is a point beyond which all that is added is sky fog which will actually decrease the signal. For my location and my equipment on an average imaging night this limit is about 600s at best, at times even 300 or 450s per sub have given good results.

Can I ask how you worked out this limit? I capture from an urban area and often thought there must be a limit to the length of a single exposure before the LP dominates the signal.

[lensman57]

Can I ask how you worked out this limit? I capture from an urban area and often thought there must be a limit to the length of a single exposure before the LP dominates the signal.

Hi,

Trial and error is for me. There is a scientific way of taking a 3 minutes exposure and reading the sky back ground value, then entering the scope focal length, dark current value of the camera and a load of other parameters into the relevant software and then it will calculate the time for you, too much science for me I am afraid. With NB filters I can double those times that I use for OSC.

Regards,

A.G

[dph1nm]

I think the term "averaging" always causes confusion, by suggesting there is something magical about the process. In fact the increase in signal to noise comes about by having more exposure time (i.e. more photons) by adding the subs. So a 3 hours exposure is just that, whether is it 1x3hr or 3x1hr or 180x1min. As others have said, having more subs add an extra noise contribution from the read noise.  This may make your final image worse, but it may not - it depends on the ratio of read noise to your other sources of noise in a sinle sub. So it is in principle possible for 600 1s exposures to give the same image as one 10min one - but you would be hard pressed to find a camera capable of doing it.

NigelM

[Spoon]

If, like an idiot, I sit outside collecting 22 hours of data one thing is for certain sure; I 'm darned well going to call that 22 hours!!!

Olly

22hours???? What object?!?

Sent from my iPhone using Tapatalk

[russellhq]

Hi,

Trial and error is for me.

Thanks A.G., though when you say trial and error, what are you actually looking for?

I've read that all you need to do is make sure the histogram "mountain" is separated from the left hand side. But surely there's a bit more to it than that?

I'm taking 900s exposures with my setup but with little clear sky time, I don't have the luxury of doing a lot of trial and error so I just went for the longest reasonable exposure time.

[lensman57]

Thanks A.G., though when you say trial and error, what are you actually looking for?

I've read that all you need to do is make sure the histogram "mountain" is separated from the left hand side. But surely there's a bit more to it than that?

I'm taking 900s exposures with my setup but with little clear sky time, I don't have the luxury of doing a lot of trial and error so I just went for the longest reasonable exposure time.

Try 300s, 450s and 600s and try and read the sky background value in your imaging software. If you are imaging from LP zone then a value of about 2000~3500 for the background is fine. For a dark  site it will be a lot lower, 900s with a fast scope and LP is a bit of an overkill.

Regards,

A.G

[asad4javed]

I am truly pleased to read this website posts which carries lots of helpful data, thanks for providing these kinds of statistics.

[wxsatuser]

Can I ask how you worked out this limit? I capture from an urban area and often thought there must be a limit to the length of a single exposure before the LP dominates the signal.

For DSLR the best way to workout your exposure for given sky conditions and camera settings, is to look at the histogram.

Take test exposures and look at the histogram, with Canon cameras the mountain of the histogram wants to be between the 20/40%

lines, that keeps the data away from the lefthand side. The peak of the mountain will come in around the 25% mark, this will give you

the best exposure for the conditions and camera settings.

Then just snap as many exposures as you can or want.

BTW

The peak is the sky background, stars etc are lowdown and to the right.

Always err on longer than shorter exposures.