Long Exp. v's loads of stacked subs?

[gary1968]

I asked this question to Blinky at the weekend and he gave his opinion, cheers Craig. I'd like to ask the same question here and get some more opinions.............

So, when imaging DSO's we need to collect loads of light. It can be done by a small numer of Long Exp's or more Shorter Exp's. The question is this, does 60 x 60 sec subs give the same final result as 120 x 30 sec subs?

Cheers,

Gary

[mostschaedel]

Hi Gary!

My personal rule for exposures of Deep Sky Object is

to expose as long as possible AND produce as many images

as possible.

But, if your object is a bright one, the 120 x 30 sec will give

better results than the 60 x 60 sec.

If your object is a very faint one it will not be detected on the

120 x 30 second images.

Generally i think the 60x60 would be better.

Much better would be 120x60sec.

[Psychobilly]

Depends on the target the camera sky conditions and a few other things ...

I like to mix 180 and 360s subs at ISO800 and 1600 unless its soemthing like M42...

1 Hour for a quick look , 2 hours getting there, 4 hours better and occasionally 8 hours...which is normally normally enough...

Peter....

[Merlin66]

Provided you can get a reasonable s/n with the shorter exposures, the stack should be equivalent.

Generally the longer the sub the better the outcome.

[Psychobilly]

Get to know how the camera performs Gary "explore" the envelope...

You shouldn't have noise issues with long exposures with the Orion... so it's more likely that you limit will be defined by the quality of the tracking...

Autoguiding is the next "logical" step... and will open up so many more opportunities...

Peter...

[gary1968]

Thanks for the replies guys. I am looking for a guide camera just now, thinking of the qhy5 but need to save a bit first.

It will be the brighter / easier dso's to start with. So am I right in thinking that exposures upto 60sec will be fine, and lots of them to stack?

I did think the limiting factor at the weekend Billy was my poor polar alignment. I need to get into WCS and drift align properly if I am to stand any chance of getting good results.

With regard to guiding, when I do get a guide camera is good polar alignment still critical? or does the fact that it is guiding reduce the importance of polar alignment?

Cheers again for the replies..............

Gary

[PAC]

I've often wondered about this one. I recently took just 2 images of M51, one for 20 mins and the other for 30 mins (both autoguided). The single 30 min frame could be stretched in photoshop to reveal background galaxies in the region of 18.6 magnitude. Would be interesting to see if 30 x 1 min subs could do the same.

Here's the single 30 min frame.

One thing to remember is that long exposures will burn out the star colours leaving them all looking white. Would be better to combine a few short and long subs together, I guess.

[MartinB]

The idea of taking long exposures is to increase signal to noise ratio. There are 2 kinds of noise - shot noise and read noise. Shot noise increases with exposure but, fortunately at a slower rate than signal. Shot noise is removed by the stacking process so, with regard to shot noise, the s/n ratio will be the same from lots of short exposures or one long one of the same equivalent time (if we don't factor in the time lost whilst reading the CCD and downloading the image).

Unfortunately read noise isn't eliminated by stacking. It is a constant amount of noise present in all images. Unlike shot noise, read noise doesn't increase with length of exposure, it is a one off hit that occurs with the reading of the CCD. The key is to have long enough sub exposures to make the read noise insignificant.

With our skies with also have to consider sky glow. Skyglow increases with exposure time and after a while any read noise is lost within the skyglow. Exposing beyond this point is of little benefit and probably detrimental.

So you should determine your sub exposure time based on the background brightness of your subs. The value of the sky glow you are aiming for is very chip dependent but for CCD users there is a guide here but you will need to know your chip characteristics

http://www.ccdware.com/resources/subexposure.cfm

Astrodon have a more user friendly RGB calculator

Astrodon Astronomy Filters - Software

[ngc2403]

hey gary,

Sorry Martin but i disagree about most of the stuff in your post...

long enposures are there to make sure we get good S/N.

read noise is a one off event

i agree, but shot noise and read noise are joined by Fixed pattern noise too. your background increases with time just like your signal, and shot noise is not removed by stacking! i only wish it were. the read noise is not the main limiting factor in the poor quality of short exposures, it's poor S/N. Stacking frames averages read noise to zero but this only helps improve the S/N after the darks, flats and most important of all for short exposures the bias frames have been subtrated off.

if you want a S/N in an image like that of one image that is 10 times longer than the one you have at the moment then you need to take 100 images. (10^2)

an earlier post asked if 30 one minute sub would be as good as 1 30 minute sub the answer no not even close TBH, the 30 one minute subs would have the same S/N as one 5 minute sub (assuming that the read noise is properly removed with darks etc) try it and see

The Fixed pattern noise limits the maxium S/N you can get with your camera and for old DSLR's it can be quiet bad. good flats remove this completely (99%)

like others said some bright objects that have well defined edges don't need the very long subs that others need to get the same sort of image at the end but that is only because the have reached a high S/N already in the short exposure so there is not need for the longer subs.

ally

[gary1968]

Thanks guys, you have given me lots to think / research about as there were quite a few 'terms' in there that I know nothing about (yet), S/N, read noise and shot noise.

So a fair bit to think about, and a lot to learn before I get reasonable results.

Cheers,

Gary

[Psychobilly]

Gary...

Don't read too much unless that's your thing... just try differnet things and see what works... some people worry far to much about the in's and out's of a ducks whatnot ... instead of getting out there and finding things out for themselves and possibly enjoying themselves in the process...

Peter...

[ngc2403]

i am stuck in the middle of glasgow 20 mile from home and my telescopes it's enough to drive me mad every time i wlak back from the library and find that there are stars over head.

FYI S/N or SNR is the signal to noise ratio which is a measure of the quality of a image.

Try this;

offset is the amount of money you have in the beginning (rainy day fund)

read noise is the small irregularites in this value with what you think it should be

Signal is your wage each month into the bank

shot noise is like standing orders, they take a fixed % of the wage

So, you get more money to spend on telescopes if you save for a longer time and you are better off (better S/N) the higher the wage you get compared to the standing orders you have.

not excatly what is happening but might help you get the idea

[Psychobilly]

I'm glad there are a few people who do read a lot and can express the concepts in laymans terms... or to whatever level of complexity they feel is neccessary or appropriate...

Peter....

[ngc2403]

only passing on knowledge that was passed on to me , but to be truthfull nothing bring this stuff home like seeing it in action

[gary1968]

Thanks ally, that does make it a bit clearer for me, I'm a complete newbie to digital imaging, not just astro imaging.

And Billy, if the bloomin' sky ever clears here I may well get the gear out into the garden and 'experiment' with different length exposures.

Cheers,

Gary

[Psychobilly]

my last but one post wasn't aimed at people like yourself NGC2403... we all need "specialists" to help us out form time to time... what i was trying to say badly is... half the fun is experimenting and finding stuff out for yourslef... I have enjoyed slaying a few sacred cows during my R&D lifestyle including modifying the design of a "long" established Dye Laser design that had really awful output stability...doing the empirical experiments and coming up with a much better design... that was stable as long as you kept the dye chemistry in check locked myself away in the lab for 2 months on that one - no one else was allowed in until the exercise was finished... and I wouldn't tell anyone what I was doing until I could prove the results as they all though at the time I was mad... perhaps they were right

For me the imaging is a hobby and getting the most of GP DSLR's is part of the challenge...

Peter...

[MartinB]

I would be very interested in reading your primary sources since it gives very different info to everything I have read. Read noise certainly isn't removed by darks and forms part of fixed pattern noise and shot noise is removed by stacking. Well it would be a sterile debate except for the fact that working out sky glow and using that to calculate sub exposure time is of such fundamental importance that people mustn;t be mislead.

Please could you give me some pointers to your sources

[ngc2403]

Most of my know is from reading the James R. Janesick book on Photon transfer called 'DN to Lambda' were he goes through the sources of all noise in CCD and CMOS cameras.

other sources include the another book by Janesick on CCD's in general, 'Scientific Charge-Coupled Devices' which includes pretty much all his knowlegde of CCDs

quote from the Net

''About the Author

James R. Janesick has written more than 75 publications on CCDs, including many NASA Tech Briefs, and holds 12 patents for various CCD innovations. During his 22 years at the Jet Propulsion Laboratory, he focused on scientific CCDs and support electronics used in many NASA spaceborne imaging systems, including the Hubble Space Telescope and the Galileo and Cassini projects. Janesick received NASA medals for Exceptional Engineering Achievement in 1982 and 1992.''

also R. Crisp of the site 'narrowbandimaging', Crisp works as a project manager for a CCD company in the US and is recommended both books, however really Janesick is the daddy of all Astro CCD's having introducted them to Astronomy

[ngc2403]

Martin,

i agree with you about the skyglow stuff, it is important to include the LP in the calculation about the exposure time.

I didn't say read noise was removed by darks only that read noise is the limiting factor after dark current, fixed pattern noise and offset, of which thes are removed by darks, flats and bias frames. read noise being random can only be averaged to zero with lots of subs.

removal of shot noise is impossible because there is no way of knowing what is signal and what is not. Hence why high quality Scientific detectors are quoted as being shot noise limited.

Perhaps a debate would be the best thing for this?

[narrowbandpaul]

a quick pointer on noise.....

there are 3 types of noise that affect all images, especially astro images, as these have low signal typically.....

noise here is just a measure of how the signal varies from pixel to pixel. ie the size of the fluctuations.

First Source of noise is read noise.....

this is infact a blanket term used to combine many different noise sources, all of which are independant of signal. That is, read noise is a constant thing. Read noise occurs from measuring how much charge was collected in the pixel. It also comes from digitising that charge...ie the ADC. If you take a bias frame, the majority of that noise is read noise....

The offset in theory is a single number added to every pixel. The fluctuations about that single number, that is noise...read noise.

Read noise limits the minimum signal you can record.

Shot Noise, or Poisson Noise.

There is a fundamental error imposed whenevr you try and count stuff like photons. This error is governed by poissonian statistics. This noise is called shot noise. It varies as the square root of signal. ie if you collect 10,000e during an exposure then the shot noise is 100e-

This noise can be reduced by stacking. Shot noise is the fundamental limit for a CCD. For a single exposure you cant do any better than shot noise. We strive for shot noise limited exposures.

Fixed Pattern Noise.

This noise will kill you.

Once charge is generated in the silicon, it needs collected by electric field supplied by the pixel. However some pixels will collect charge more efficiently than others. This leads to a fixed pattern of more responsive and less responsive pixels....any variation in charge is defined as noise. So this collection efficiency is called Fixed Pattern Noise.

The reason that is so bad, is that it scales linearly with noise. ie, once you are in the fixed pattern noise regime, your signal to noise is a constant. It doesnt matter if you give it more subs or longer exposures. It is a waste of time.

Fortunately a flat field will remove this noise very nicely and return your sensor to shot noise limited performance. Which is total awesome.

Your flats need to be of a high enough signal level to show FPN otherwise it wont be corrected for. I would say that 20 flats averaged each at 50% full well would be a good start.

Read noise can be reduced by averaging frames, shot noise can be reduced by averaging frames, and fixed pattern noise can be almost removed entirely by flat fielding.

Removing noise sources is fundamental to improving the signal to noise ratio.

Remember we like signal, we hate noise.

Hope this was useful

Paul

[MartinB]

These are very good sources. I went to Richard Crisps site but couldn't find anything relevant about shot noise. I scan read Janesik's article there but it didn't seem to shed any light on stacking and shot noise. I might well have missed it though. If Richard Crisp or James Janesik are saying different things to the info posted below I would really like to know since it might make me think again about individual sub exposure times.

Here is a link to an excellent article by Ron Wodaski

Astromart Articles - CCD Cameras and Signal to Noise Ratio

And here is an excellent article on noise from the QSI web site. it relates to read noise but gives an explanation of shot noise

Understanding CCD Read Noise

More info here

Techniques page

On an individual frame there is no way of knowing what is noise and what is signal. Long exposures work by the fact that noise builds at a slower rate than signal so S/N improves. By stacking frames the random nature of noise allows it to be eliminated by averaging.

The ability to stack frames is massively important not only for people with light polluted skies but for when imaging when there is the odd scudding cloud, gusts of wind and aircraft to think about.

[MartinB]

Gary, I'm sorry we've drifted off into the detail of noise. I appreciate that you don't need to know the theory of the internal combustion engine to drive down to the supermarket! As you will have seen, the answer to your question involves all sorts of considerations.

My advice would be to expose up to a point where you are starting to get sky glow affecting the image. Then collect lots of exposures of this length and stack them.

[roundycat]

Gary, I've so far kept out of this one and there has been some good and some not so good advice given. As Martin said above everyone has wandered off the subject a bit as well. There is nothing I like more than a nice theoretical fireside chat about something obscure and this subject certainly qualifies. The thing in astro imaging though is to keep your feet where they belong, firmly on the ground. Post#9 above talks about subtracting darks, flats and bias which really is confusing. Rather than bash off quick replies it would be more constructive if people would think before posting. Darks contain bias so you do not subtract both, flats are not subtracted anyway.

How about a little practical help in the form of a direct comparison. Anyone can do this and perhaps some should so we can compare different techniques and different results. Fighting with thick text books is ok insofar as it goes but we all know where the real proof of the pudding lies. As an aside, it seems to me that there are some who can ruin a stack of good exposures simply by using the wrong technique or software. Some give me the impression they have been stacked with a coal shovel.

Shown below are two similar sized shots of the Rosette. One is 5m and the other 120m. With suitable (or should that be unsuitable?) contrast mods in PS and smoothing with Neat Image or similar they can be made to look nigh on the same.

The pursuit of noise free pics is an honourable one but remember how the pics are viewed, on a shaky old monitor or a print that is being viewed in unsuitable light. As someone said above get out and play with your kit and see what happens.

Dennis

[roundycat]

here is the pic

[CELESCOPE]

Excellent Dennis , been interested myself with this thread, being here in the LP city lol , i seem to be ok with Nebs ,but with taking on Galaxies now with a CLS filter , things have become worse. my very words to Dave this morning , whats best short of long subs , i think as you say we can all get lost in this text book stuff, not downing it in anyway ,but just like you have done, Dave suggested , ,go out and play take different exposure lengths and find the best that suites the Lp and equipment we are using ,as this will differ for everyone.

So thats what i will do, on the next clear night.

Cheers

Rog:)