Concerning Exposure

[oopfan]

You may not know this but I've been a strong proponent of long exposures, much to Robin's chagrin. Since then I've moderated my stance after developing a calculator and testing it in real life. What it tells me, first and foremost, is that my Bortle class determines my maximum exposure. Of course, I can go longer but I don't gain much. In fact all I do is increase the risk of star saturation. On the other hand there is such a thing as an exposure that is too short.

So what do I mean? With my calculator I play this game of "doubles and halves". I start with a guess: a 30-second exposure. Then I ask "how much better will my image be if I double the exposure?" If the answer is "a lot" then I make my new exposure 60 seconds. Then I repeat the process. On the next iteration to 120 the calculator might say "ahh, sorry you aren't gaining much." I also run it in reverse by starting with 30 seconds and then halving it to 15 seconds. If the answer is "stop, you will need to dramatically increase your total integration time" then I stop and call 30 seconds my minimum exposure.

You wind up with a range of acceptable exposures, say from 30 seconds to 60 seconds. The next question is "what does my field of view look like?" What you do depends on your answer. If there is a bright star then I recommend using the 30 second exposure but if your field contains lots of faint stars then I say go with 60 seconds. SharpCap's Smart Histogram/Brain takes care of this decision-making for you except for that last part about what your field of view looks like. My recommendation (subject to Robin's approval) is to halve the exposure if your field contains bright stars.

An example follows...

Look at the attached side-by-side image and tell me which is better? (OK, they are both crappy but that is besides the point.) If you say the right image, then what conclusion can you draw? (Look at the fine print.) The right image used 90 seconds exposures and the left used 30 seconds. Are you saying that the right image is better because it used 90-second exposures? Look at the overall picture. How about those stars? There are a lot of them that are fat indicating extensive star saturation. Now look at the left image. The stars look a lot better but the nebula is very noisy. Why is it noisy? It is noisier because it is fainter! In fact I'm saying that if you doubled the exposure then the nebula would not look any better but the stars would be fatter! Annoyingly so. The only thing that can help the image is more integration time.

Just a sidebar, I have this rule of thumb called the "1000 ADU Rule." (It is based on the Optimal Exposure Calculator available at Starizona's website.) What it says is this. For a given exposure if the image background is less than 1000 ADU then increase the exposure, but if the background is more than 1000 ADU then decrease it. What this implies is if you move to a dark site, say Bortle 3, then it will take a much longer exposure to reach 1000 ADU but if you are in the city at Bortle 8 then it takes only a few seconds.

Having said that, to prove my point, the sky background of the right image was as high as 10,000 ADU but the image on the left is at 2,000 ADU. Again this just proves that the left image is optimal even though the nebula is noisy. That is because it is faint. The only remedy is integration time

LDN-1235_compare_LDN-1251.png (599.49 KiB) Viewed 1906 times

Brian

[Menno555]

Thanks for this Brian.
There is indeed a kind of logic to this which, like many times is said, also depends on the whole picture: Bortle 2 or 8, equipment, filters, combination of focal length and camera (over- or undersampling), and so on.
For my f/10 scope that part of a bright star being in the FOV is all saying. Like you I start with a certain exposure that "feels" right and (depending on the object) and then go up in exposure time until that star starts to bleed. At that point it a notch down in exposure time and that's it.
And if there is no bright star in the FOV, then indeed it's my Bortle 8. Any longer then 180 seconds and I'm in trouble.
So for me it's short exposures and a lot of them

Having said that: if I would be at a dark spot and no bright star in FOV, I would go for longer exposures for sure. Fatter stars can be dealt with in processing. And from what I've seen, the longer exposures just show (way) more detail, especially with fast systems.
But when it comes to short exposures compared to long exposures, also here it seems on the web that there are 2 sides "fighting" each other. A bit like the battle between SCT and refractor users
For me it's simple: use what works best for you. If I had to choose between the left or right picture you are showing, for me the right one is the one to go with. Partly because I don't know how to measure ADU for a background and partly pure on sight.

Btw: that Starizone site is not really optimized good somehow. On my Chromium browser it only shows a bar at the top and then an empty page. In Edge and IE Explorer the page is showing and I can enter values. But when the Calculate button is pressed, nothing happens. The button turns blue but no value is shown.

[oopfan]

Hi Menno,

Regarding Starizona's calculator, two years ago it was working. I peeked at their JavaScript code that runs in the browser and got the gist of what they were doing. If I remember correctly their instructions called for capturing a frame, calibrating it, and then measuring the number of ADU counts within a rectangle of dark sky. Frankly, I think that performing calibration is overkill as long as you know what your mean bias level is. For my Atik 314E it is 220 ADU. So then all I do is sample the ADU counts in an area of "dark sky" and subtract 220 in my head. If it is less than 1000 then I increase exposure. Some CCD's that I use have a bias of 2000 ADU so I am looking to find an exposure that puts "dark sky" above 3000 ADU.

With regards to which of my images in my initial post are best, it is subjective but I tend to like images that have "balance" and that reflect reality as much as possible given my equipment. In the right-hand image I agree that the abundance and intensity of stars is impressive but what do I want the user to see first and foremost? The nebula. The stars play a secondary role but I don't want them to distract from the true "star of the show", the nebula.

Regarding the use of software to reduce the size of saturated stars, I think that technique should be applied sparingly. There are extreme cases like Alnitak, the bright star next to the Horsehead Nebula, that needs to be tamed. My other interest is Photometry so I am always concerned about avoiding star saturation. I apply similar techniques to Astrophotography where I set the exposure so that the brightest and faintest star fall within the linear range of the sensor. That is not always achievable given the wide dynamic range of star magnitudes. In almost all cases you can be guaranteed that your field will contain a bright 8th or 9th magnitude star. It is not possible to find an exposure that also captures an 18th magnitude star without saturating the bright stars. Very expensive cameras can better handle these extremes but many times the cost is prohibitive to the amateur. So what I do is accept some saturation of the brightest stars but no more than is necessary.

One final thought on the 1000 ADU rule, many years ago when I was doing landscape photography with film-based cameras, the general rule was "don't let your shadows go dark". In other words, don't make the exposure so short that you lose detail in your subject's shadows. If you don't capture enough photons then shadows become "black holes". It is better to lengthen the exposure at the expense of over-exposing the highlights. In the darkroom there are techniques for taming highlights and thus preserving detail in them. There are no techniques for extracting detail in shadow that doesn't exist. Of course you can go overboard. I suggest that I went overboard with my right-hand image by selecting an exposure that made the sky background 10,000 ADU. This is unnecessary and it just causes greater star saturation.

Thanks for your reply,
Brian

[oopfan]

Menno,

Here is a wide-field IOTD shot of an area in Cassiopeia that I am using for planning my own effort. The red frame is the FOV of my camera. Notice the dark nebula and the small rectangle within it. This is where I want to sample when determining exposure. This is a long integration time effort at Bortle 5 so I won't see the clear delineation between Milky Way and dark nebula in a single frame. I will just see a "snake-like" void of stars. So in this case I want to sample the void. You can also use the "Brain", just have it sample the void.

Brian

Wide-Field-1-Exposure.jpg (440.35 KiB) Viewed 1816 times

[turfpit]

Here are some examples of annoying stars I have come across:

14Vul when capturing M27 https://www.astrobin.com/1graw1/?nc=user

52Cyg in the Western Veil https://www.astrobin.com/44bc0e/B/?nc=user

Alnitak just runs riot with the exposure needed to get the faint detail https://www.astrobin.com/br3na0/F/?nc=user

Deneb when chasing NGC7000 https://www.astrobin.com/rafelk/0/

I mitigated Deneb by using HDR techniques https://www.astrobin.com/zj4r5b/?nc=user, maybe some shorter exposures would help here.

My understanding is that well depth of the camera will help with star saturation. I think my Atik314E comes in at 13,000. Ideally I would like to get a camera which gives me a well depth of 50,000 - 100,000. Sadly the depth of my wallet cannot support the camera well depth I aspire to.

Some expansion on the mean bias ADU and the excess of at least 1000 in an area of dark sky that Brian mentioned. I have included below examples of how I find this using FITS Liberator and Siril for an Atik314E and an Altair183C. A dark area mean ADU of a light frame is included at the end.

Atik314E-bias-mean-ADU-FITS-Liberator.jpg (108.4 KiB) Viewed 1813 times

Atik314E-bias-mean-ADU-Siril.jpg (86.92 KiB) Viewed 1813 times

Altair183C-bias-mean-ADU-FITS-Liberator.JPG (70.76 KiB) Viewed 1813 times

Altair183C-bias-mean-ADU-Siril.jpg (49.87 KiB) Viewed 1813 times

mean-ADU-in-dark-area-of-a-light-frame.jpg (27.19 KiB) Viewed 1813 times

The last image is a light frame, showing a dark background area selected. For the 183C, Mean ADU for the light frame = 1999.5. The mean ADU for the camera's bias frame = 451.1. The difference is > 1000 and therefore I am happy that the exposure is sufficient.

YMMV.

Dave

[oopfan]

Dave,

Thanks, great screenshots with explanation. I too use Siril for this purpose. There are distributions for Windows, and it is free. Just to be clear you can also use the "Brain".

Brian

[Menno555]

Menno,

Here is a wide-field IOTD shot of an area in Cassiopeia that I am using for planning my own effort. The red frame is the FOV of my camera. Notice the dark nebula and the small rectangle within it. This is where I want to sample when determining exposure. This is a long integration time effort at Bortle 5 so I won't see the clear delineation between Milky Way and dark nebula in a single frame. I will just see a "snake-like" void of stars. So in this case I want to sample the void. You can also use the "Brain", just have it sample the void.

Brian

Ah okay! I've been taking the instructions for the brian ... eh sorry: brain to literal then. I used it till now always as described: a dark patch without many stars or nebulosity. Thanks for the tip!

Menno

[umasscrew39]

Very interesting discussion and viewpoints by all. Over time, I've learned to start checking for star saturation prior to image stretching to see if I am overexposing via PixInsight, as it has a nice way to determine this. On a scale of 0 (pure dark) to 1 (pure white), it allows you to visually inspect any star. This is a normalized scale which represents the actual pixel values from the large dynamic range rather than reading 0 to 65535 or from -3.4E+38 to +3.4E+38, the range for 32 bit float numbers.. An accepted rule is that any value above 0.8 is representing saturation which would indicate star bloating and may even begin to distort star color, which only gets exacerbated upon stretching to bring out the DSO. Other variables which also influence this are scope setup, filters, and sky conditions. So the solution is to go to shorter exposures as was nicely explained by Brian.

Here is an example of a star reaching near saturation from an image of mine of the Needle Galaxy.

Screenshot (6).png (132.85 KiB) Viewed 1778 times

[oopfan]

Hi Bruce,

High-end cameras have a Full Well Depth (FWD) of 100,000 electrons and Anti-Blooming Gates (ABG) that facilitate long exposures (~30 minutes) at dark sites. My CCD has a FWD of 13,000 electrons so all I have to do is breathe on it and it saturates so I need to be careful choosing an exposure. At my Bortle 5 site a 60-second exposure is pretty much ideal. Total Integration Time decreases only marginally if I increase the exposure but if I lower the exposure to 30 seconds then my integration time starts to really increase. Yet still at 60 seconds my camera will saturate 9th magnitude stars, and there are plenty of those in each field of view! The only cure is a more expensive camera. Fortunately for photometry I can effectively double, triple, and quadruple my FWD by de-focusing. That little trick doesn't help with AP though.

Thanks for your reply,
Brian

[umasscrew39]

I guess my point was while not a cure-all, this simple approach that PixInsight provides does help one to get an initial sense of star saturation regardless of the CCD or CMOS being used for AP. The "fixes" are many from prevention to post-image adjustments.