New Forest Observatory ®

I have just seen a bunch of YouTube videos on this subject, and they are all pretty confusing and worse still, contradictory. The issue of sub-exposure time (and the number of subs) is something I have worked on with Noel Carboni for something like 20 years. During that time I have oscillated between the idea of LOTS of short subs, and using a smaller number of long subs. I have now settled on Noel’s solution as being the best, it is when you think about it, completely obvious. But before giving you the Noel Carboni answer, let’s look at some of the things that should concern you regarding sub-exposure time. If you can’t be bothered with all the rubbish below – then go straight to the end for the Noel Carboni (definitive) answer.

See if you can find and read https://www.newforestobservatory.com/2021/10/08/a-10-year-journey-on-speed/ where you will find a lot of what I am about to say later in this article.

The sub-exposure time I am talking about here IS NOT the “optimal” exposure time (which from my experience is anything but optimal) but it is the exposure time you should be using in your location, with your sky background, for the DEEPEST/HIGHEST SNR (signal to noise ratio) image you can get.

Let’s kick off with the f# of your system and see where that takes us. f# (focal length of system/system aperture) REALLY IS the speed of your system. So my f#2 Hyperstar (on a C11) is 5x faster than my Sky90 at f#4.5 (with its reducer/corrector). So to get the same DEPTH of image, if I take a 1-minute exposure on the Hyperstar, I will need to take a 5-minute exposure on the Sky90 – nothing contentious there. I guess I should also add that I work under Bortle 4.5 skies (Bortle 5 towards the East and Bortle 4 towards the West) so I am not talking about those dreadful Bortle 6 and above skies that some of your poor Devils deal with, they need completely different approaches. The same goes the other way of course. I am highly envious of those that have Bortle 2 or even Bortle 1 skies who on a clear Moonless night could take 20 or even 30-minute subs with an f#2 Hyperstar for breathtaking results – but I begin to digress. In my early days of imaging I used to take 30 – 60-second images with the Hyperstar, and lots of them! Why? Because I didn’t know what I was doing and because the glassy smooth results (high SNR) looked good to me. Also, and I didn’t know this at the time either, the stars came out pretty good as well. Nicely round and quite small. With long exposures you are creasing yourself if you want nice small, round stars, as you will not only get some “bloating”, but you are also testing the polar alignment/tracking capabilties of your mount AS WELL. So the WHOLE story, is as you can see, already becoming a bit complicated! Never mind, we’ll plough on. So I took an f#2 Hyperstar image of the Merope region in the Pleiades, something like 100 x 30-second subs, and sent it off to Ron Arbour to identify an asteroid going through the Merope nebulosity (which he did). The image looked great to me, glassy smooth (that’s down to the 100 subs) and plenty of blue nebulosity. Ron made the passing comment that it was indeed very high SNR, but it didn’t go very deep! I didn’t understand what he meant by this until I did a lot of work on the Pleiades with the Sky90 array and much longer sub-exposure times (and MANY more of them). It was only with the Sky90 data that I began to see the (brown) Taurus molecular cloud permeating the Pleiades, something completely missing from the Hyperstar data. The upshot of all this is that 30-second subs are not long enough under Bortle 4.5 skies for the Hyperstar at f#2 to get hold of the faint stuff. If I can get deeper results (showing the TMC) with longer sub-exposure times on the Sky90s, then I am clearly “creasing” myself by taking only 30-second subs on the Hyperstar. There is yet another proviso (of course). f#2 is SO FAST that you are easily creased by poor skyglow, which means if the biggest light polluter in the sky is up (the Moon) – you’re better off giving Hyperstar imaging a miss and moving over to a slower rig, if you have one.

So 30-seconds at f#2 is too short under Bortle 4.5 skies – at least we have a lower bound. So what’s the upper bound? That’s a tough one, and it depends on the object you are imaging, and what the sky is like in the region you are imaging the object. If you are imaging faint stuff, at the Zenith then you can get away with a longer sub-exposure time than if you are trying to image faint stuff down in the murk low on your southern horizon. And yes, I still haven’t given any hard numbers yet, sorry about that, but as you can see, it is a messy business. If I have a bright object filling up most of the field of view (FOV), something like M31, then 3-minutes is absolutely fine. Get 100 x 3-minute subs (only 5-hours!) of M31 and if your focus and collimation are spot on, you will end up with a really first class image. Go for something a little less bright, but also filling the FOV, like the Rosette nebula, and I find 5-minute subs are appropriate. Go for something even fainter like the North America nebula and I can go up to 10-minute subs with nice results. Finally, I have even gone to 15-minute subs with the Iris nebula region on a good night with great results. There are of course negative aspects with going to the longer subs – it becomes more painful to lose a sub! Like everywhere else on Earth I am plagued by satellites – but they can be dealt with in the stacking software – I am also plagued by damned planes – and they simply wreck the sub. You also increase the risk of losing a sub by a “jogged” frame, however that occurs. O.K. so now we’ve found the (rather huge) range of sub-exposures for f#2 imaging – how many of them do we need for a decent SNR? Empirically I have found that from my Bortle 4.5 skies if you can take in excess of 8-hours of data you should be in good form. So that’s 32 x 15-minute subs, 48 x 10-minute subs, and 96 x 5-minute subs. 8-hours imaging also means, unless you live in a nicer place than me, multiple nights of imaging, typically 3 nights. I have not mentioned anything about filters on the Hyperstar which would allow you to push up the sub-exposure times, for one reason only. I have not found, and I do not believe they exist, ANY filter that is suitable for the Hyperstar at f#2. If you know anything about optics and interference filters it is obvious why this is so, but if you believe you have found such a filter, good for you!

What about sub-exposure times with an f#4.5 refractor? These are of course going to be longer (typically) than for the Hyperstar, and as you need quite a few of them as well, you can see why I went for an array of 3 x Sky90s so I can get down 3-hours of data in just 1-hour of actual imaging time. So for those very bright main constellation stars that I have with the star at centre, you can’t go to very long subs without flaring everything out. So for those you are forced to go to short 2 or 3-minute subs. This means if you want to get any low brightness stars in the region you need to get LOTS of subs (to drag them out of the noise) and this in turn means getting around 4-hours or more of data (so that’s just over an hour on the array!). For typical star fields I typically use 10-minute subs and like to get around 50 subs for a high quality image, so that’s around 3-hours only on the array, but a huge 9-hours on a single Sky90. I typically use 15 – 20-minute subs on fainter stuff like nebulae, so the numbers start sky-rocketing even when you have an array, and I have easily spent in excess of 12 imaging hours (that’s 36 hours of data) on many nebulae.

So then, after all that, what IS the answer? It’s very simple. You get as many sub-exposures at the maximum sub-exposure time for your particualr object, under your sky conditions (at the time), with the f# of your system. Before an imaging session with the array I will eyeball the region of sky I want to image in, just to get an idea of the maximum sub length I’ll get away with. On a poor/average night that might mean only 10-minute on the Sky90s, but with ideal conditions it could be 20 or even 30-minutes per sub. There is NO one answer, and it depends on ALL the stuff I’ve mentioned above.

POSTSCRIPT: THEORETICALLY the main thrust of the article I have written above is complete rubbish. Very many astrophotographers out there will say you can get images as DEEP as you like with high SNR, with even SHORT subs, so long as you get enough of them. This is particularly espoused by those poor souls working with terrible Bortle skies – I guess it gives them hope. Practically I’m afraid this is complete nonsense (don’t ask me why, I’m no good at theory). So I offer you a challenge here to prove me wrong. Show me any image of an object I show on this site, taken with short (say 1-minute or less) subs (under poor skies) that even begins to approach the depth and quality of my images. Why do I know you can’t? That’s very simple – it’s because even my best images look equally crap compared to those taken under Bortle 1 or 2 skies – and there ain’t nothing you can do about that.