Chris, Ray has produced a very useful chart that shows optimal broadband sub times for differing Gain settings , for three different f ratios and two different levels of darkness (SQM), it is published in the ASI1600 thread here.
Thanks Glen, I have seen that chart (and will use it for now), but I wanted to understand more, plus be able to do rough calculations quickly for other cameras and situations.
Quote:
Originally Posted by glend
I would suggest eliminating variables such as moon effect, and time of night light effects ( like nearby sports fields, neighbors, etc).
Of course narrowband is not as suscepable to 'external influences' imho.
So, you're suggesting finding a time to measure the best conditions then? New moon (or no moon around) night, neighbours asleep with lights off, etc.
This will then lead to finding the time to being sky limited under the best conditions, which will be longer than in most times of the night/month (suboptimal conditions). If you're taking exposures over a number of hours or nights, you're likely to keep T consistent (for darks, stacking, etc). Therefore, if I understand correctly, a proportion of those exposures are going to be longer than the current sky limited value at the time of being taken. And this in turn should mean that it's ok to go above this time, but you're not going to gain anything extra (you're just wasting a bit of time).
It's interesting to see this discussion on theoretical SNR....
I doubt many AP imagers actually measure the SNR in their images.
In spectroscopy we must measure the actual SNR in our data to comply to the ProAm campaign requirements.
We usually have to target SNR>200
The shot noise contribution is not IMHO affected by the use and summing of subs. The total signal determines the shot noise.
Data of 100,000 photons will have a shot SNR of 316 whether it came from 1 or ten subs.
Also remember that ANY manipulation of the data will add noise. This goes for darks, flats, colour balance corrections etc. etc. etc.
In the SimSpec spreadsheet we use to analyse the performance of the spectrograph Christian Buil presented a method of calculating the anticipated SNR which includes all the factors discussed.
(See also: Howell's "Handbook of CCD Astronomy", Sect. 4.4, p73 gives the "CCD Equation", Budding & Demircan's "Introduction to Astronomical Photometry, Sect 5.2.6 - Noise in photometry, p188, Appenzeller's "Introduction to Astronomical Spectroscopy", Sect. 3.4.2, p78)
(The SNR comments above are based on the original post:
3) All else being equal, then for the very faintest targets, the snr is proportional to exposure time T, but proportional only to the square root of the number of subs N. That means that for very faint targets, it will be hugely better to do ten 3600 second subs than 36,000 one-second subs. It will be better by a ratio of 3600 root(10) to 1 root(36000), or a ratio of six to one. Long subs rule.)
...................
3) All else being equal, then for the very faintest targets, the snr is proportional to exposure time T, but proportional only to the square root of the number of subs N. That means that for very faint targets, it will be hugely better to do ten 3600 second subs than 36,000 one-second subs. It will be better by a ratio of 3600 root(10) to 1 root(36000), or a ratio of six to one. Long subs rule.)
I have nothing useful to add to this other than: amen
... The shot noise contribution is not IMHO affected by the use and summing of subs. The total signal determines the shot noise.
Data of 100,000 photons will have a shot SNR of 316 whether it came from 1 or ten subs. ...
Totally agree. Never said anything to the contrary. Completely obvious.
Quote:
Originally Posted by Merlin66
... (The SNR comments above are based on the original post by Placidus:
3) All else being equal, then for the very faintest targets, the snr is proportional to exposure time T, but proportional only to the square root of the number of subs N. That means that for very faint targets, it will be hugely better to do ten 3600 second subs than 36,000 one-second subs. It will be better by a ratio of 3600 root(10) to 1 root(36000), or a ratio of six to one. Long subs rule.)
Merlin, are you agreeing or disagreeing with my statement 3?
You seem to be disagreeing on the grounds that the partitioning of shot noise into many short subs does not affect the signal to noise ratio. That is like saying that since tables have four legs (true), they cannot be used for serving food. Your true statement does not disprove my true statement.
Partitioning the same total exposure into a large number of subs increases the snr by adding many doses of read noise instead of just a few doses. The equation that I gave (for the case of no dark current noise and no sky noise) demonstrates that. It is still the case with the more complex equation that adds in the role of dark current noise and sky noise, although the effect is less or much less in sky limited shots.
3) All else being equal, then for the very faintest targets, the snr is proportional to exposure time T, but proportional only to the square root of the number of subs N. That means that for very faint targets, it will be hugely better to do ten 3600 second subs than 36,000 one-second subs. It will be better by a ratio of 3600 root(10) to 1 root(36000), or a ratio of six to one. Long subs rule.)
However, 288000 one second subs would be equivalent, right? If you can't take long subs for whatever reason, just do more. It will just take longer overall and what's wrong with that? Really, long subs only rule if you count elapsed exposure time (and also assume zero failures in the capture process).
I'd prefer to get back to the practical issues for each person. How does one determine the best for their own setup. And if possible, I'd really like some of my questions from earlier posts explicitly answered. :p
Guys,
I was referring to shot noise and shot noise only....
The quoted original para 3 doesn't say that the difference in SNR is due to the varying shot noise....later quote: "by adding many doses of read noise"
Theory is great, but what about the noise you add to your images through processing??????
Lazjen,
I'd love to help if you can explain how you currently measure the SNR in your images....what "improvements" are you looking for????
Guys,
I was referring to shot noise and shot noise only....
The quoted original para 3 doesn't say that the difference in SNR is due to the varying shot noise....later quote: "by adding many doses of read noise"
Theory is great, but what about the noise you add to your images through processing??????
Lazjen,
I'd love to help if you can explain how you currently measure the SNR in your images....what "improvements" are you looking for????
Ken, our measure of "SNR" is the perceived graininess in an image that may be arbitrarily stretched and possibly smoothed or sharpened. There is no maths for "graininess" in this environment so we use the well developed physics of signal and noise in a linear imaging system as a surrogate - on the simple basis that, if we can get better true SNR in the underlying linear image, we will have better perceptual graininess in the pretty picture.
Using SNR theory enables us to optimise our equipment in a systematic way and relate problems in the "pretty picture" domain to the underlying physics. We do not need to measure true SNR to do this (although some of us do when pushing the limits or testing our understanding of the theory), but that doesn't invalidate the approach.
I'd love to help if you can explain how you currently measure the SNR in your images....what "improvements" are you looking for????
I'm looking for the starting point. As in, I make a decision to go for target X from location L using gear G (camera with specific settings, mount capable of handling M minute exposures consistently).
I need to answer questions like this:
* I can do this image over multiple nights, what's the best T value I can use to get the maximum theoretical SNR in the result?
* The weather is looking like rubbish and this night is likely it, or I'm at a dark site for just a night. Given I liked 20+ images per channel to process I want to take (and therefore constrain T), what camera settings do I use to get maximum theoretical SNR? This can also help determine if target X is actually feasible or I should pick something different.
In both cases, T has a max value of either M or the Sky Limited time at L (SL), whichever is smaller (actually I suppose there's also a limit based on the camera's capacity). Although I assume if SL < M, going to T = M won't hurt, but is just a waste of the extra time? This is one of the questions I'd like to know, because if I work out that for a number of common situations, that I should use a value of say 30 - 75 seconds, I'll just pick 60 sec to cover those cases, etc. If it's 200 to 280, I'll pick 300 sec just for the simplicity.
Is this all obvious to everyone else and I'm missing something basic here?
Ray,
I understand the "pretty pictures"......
The graininess of your image can actually be measured to give a good estimate of SNR using the processing software or Image J.
It could be of interest to compare perceived graininess it's the measured SRN.
I'm looking for the starting point. As in, I make a decision to go for target X from location L using gear G (camera with specific settings, mount capable of handling M minute exposures consistently).
I need to answer questions like this:
* I can do this image over multiple nights, what's the best T value I can use to get the maximum theoretical SNR in the result?
* The weather is looking like rubbish and this night is likely it, or I'm at a dark site for just a night. Given I liked 20+ images per channel to process I want to take (and therefore constrain T), what camera settings do I use to get maximum theoretical SNR? This can also help determine if target X is actually feasible or I should pick something different.
In both cases, T has a max value of either M or the Sky Limited time at L (SL), whichever is smaller (actually I suppose there's also a limit based on the camera's capacity). Although I assume if SL < M, going to T = M won't hurt, but is just a waste of the extra time? This is one of the questions I'd like to know, because if I work out that for a number of common situations, that I should use a value of say 30 - 75 seconds, I'll just pick 60 sec to cover those cases, etc. If it's 200 to 280, I'll pick 300 sec just for the simplicity.
Is this all obvious to everyone else and I'm missing something basic here?
I'll (re)read it. I think it will help. I think I need to go collect some data and check the results to make these equations and numbers more real to me.
I'll (re)read it. I think it will help. I think I need to go collect some data and check the results to make these equations and numbers more real to me.
We thought it was time to do some actual calculations with a real galaxy, (a) using a luminance filter, and (b) using an H-alpha filter.
The finished images (not the raw data) can be seen on Ice In Space here.
The results are extremely illuminating.
For a luminance shot, the sky glow is the overwhelming contributor to the noise, far exceeding dark current or in particular read noise. Consequently, 130 six-minute subs with our gear would be almost as good as 13 one-hour subs.
For a 3nM h-alpha filter on this faint target, the results were completely different. The snr for 130 six-minute shots was 1.2, and leaped to 2.6 with 13 one-hour subs. That vindicates our practice of using long subs with faint narrowband objects.
We examined theoretically what would happen with a zero readout noise camera. It would produce almost no improvement with a luminance filter, because of the overwhelming effect of sky noise. Conversely, with an h-alpha filter, it would produce about a 30% improvement for a 1-hour sub, (not enough to justify our throwing out our 16803) and a 280% improvement for a 6 minute sub, pretty enticing for those who need short subs for practical reasons.
Our grand conclusion is that one has to actually make the measurements and do the arithmetic for the specific case (equipment, location, subject) in mind.
A first draft of the detailed methods, results, and discussion paper is attached. There will undoubtedly be mistakes, even howlers. The most recent version can be seen at DropBox here.
From the recent fascinating discussions and explanations, I have finally arrived to a well-known fact, that the latest CMOS-based cameras allow many more people interested in astrophotography to actually dip their feet into this fascinating hobby due to a significantly lower cost of the camera and a much lesser need for accurate guiding. This is wonderful. It appears that CCD-based cameras will continue, at least for a few more years, to provide a greater capability to produce superior results in DSO imaging, but at a higher cost and by putting significantly higher demands for accurate guiding.
From the recent fascinating discussions and explanations, I have finally arrived to a well-known fact, that the latest CMOS-based cameras allow many more people interested in astrophotography to actually dip their feet into this fascinating hobby due to a significantly lower cost of the camera and a much lesser need for accurate guiding. This is wonderful. It appears that CCD-based cameras will continue, at least for a few more years, to provide a greater capability to produce superior results in DSO imaging, but at a higher cost and by putting significantly higher demands for accurate guiding.
I'm not sure how having a crook mount helps regardless of the camera being used...in fact...I'd preach just the opposite. Your results will be even better if the incoming flux accurately stays in the one spot on a sensor.
I'd suggest newbies buy the best mount they can reasonably afford....and only then think about the camera/scope.
We thought it was time to do some actual calculations with a real galaxy, (a) using a luminance filter, and (b) using an H-alpha filter.
The finished images (not the raw data) can be seen on Ice In Space here.
The results are extremely illuminating.
For a luminance shot, the sky glow is the overwhelming contributor to the noise, far exceeding dark current or in particular read noise. Consequently, 130 six-minute subs with our gear would be almost as good as 13 one-hour subs.
For a 3nM h-alpha filter on this faint target, the results were completely different. The snr for 130 six-minute shots was 1.2, and leaped to 2.6 with 13 one-hour subs. That vindicates our practice of using long subs with faint narrowband objects.
We examined theoretically what would happen with a zero readout noise camera. It would produce almost no improvement with a luminance filter, because of the overwhelming effect of sky noise. Conversely, with an h-alpha filter, it would produce about a 30% improvement for a 1-hour sub, (not enough to justify our throwing out our 16803) and a 280% improvement for a 6 minute sub, pretty enticing for those who need short subs for practical reasons.
Our grand conclusion is that one has to actually make the measurements and do the arithmetic for the specific case (equipment, location, subject) in mind.
A first draft of the detailed methods, results, and discussion paper is attached. There will undoubtedly be mistakes, even howlers. The most recent version can be seen at DropBox here.
Thanks folks, lovely to see some real data on what sky-limited imaging means in practice.
your conclusion that 6 minute subs do as well as 1 hour subs on your system flies completely in the face of much accepted wisdom, but it agrees with the theory - nice to see. If you attempt to scale your results to incorporate a 2e RN camera, you will see why the users of the new CMOS chips only need 30-60 second subs - there is just no point in going any longer.
I'm not sure how having a crook mount helps regardless of the camera being used...in fact...I'd preach just the opposite. Your results will be even better if the incoming flux accurately stays in the one spot on a sensor.
I'd suggest newbies buy the best mount they can reasonably afford....and only then think about the camera/scope.
I think you've taken Suavi's statement and twisted it a bit. I can't see that he's proposing anything against the common wisdom of buying the best mount you can afford.
What it means though is if someone comes to the hobby with a budget of X, that because this camera is very capable and relatively inexpensive to the traditional CCD offerings, a higher percentage of X can be spent on the mount than might have done otherwise.
I think you've taken Suavi's statement and twisted it a bit. I can't see that he's proposing anything against the common wisdom of buying the best mount you can afford.
What it means though is if someone comes to the hobby with a budget of X, that because this camera is very capable and relatively inexpensive to the traditional CCD offerings, a higher percentage of X can be spent on the mount than might have done otherwise.