3.24
#...
·
·
1
like
|
|---|
| It seems like most folks are using 5 minute exposures when shooting with LRGB or Ha OIII SII filters. Why is that? Why not use 10 or 20 minute exposures? Please educate me, I'm new to CMOS! |
6.89
#...
·
·
3
likes
|
|---|
|
CMOS tends to have lower read noise and higher dark current than CCD so it makes sense to use relatively shorter subs. Of course it all depends on the exact circumstances. Even 5 minutes could be a lot for LRGB (especially L) in the presence of high light pollution or bright stars. I don't think there is a standard. |
1.81
#...
·
·
3
likes
|
|---|
| With CMOS cameras read noise is very low and you do not need long exposures to swamp it. 4 or 5 minutes is usually long enough . This is a kind of diminishing return point for narrowband, for LRGB even shorter. With shorter exposures, the demand on guiding is less strict, satellite , airplane ,... trails are removed easier, cloud passing or wind gust does not destroy 20 minutes, it destroys only 5 or 10. Of course, the price you pay is high number of files, requiring disk space and computing power to stack. If your guiding is perfect, the weather is perfect , you have excellent dark frames library , nothing stops you to take long exposures. Of course you have to be extremely careful not to "burn" stars and intense object, as that happens with much shorter exposures than for example with CCDs |
1.20
#...
·
·
4
likes
|
|---|
|
There's a few different reasons for choosing exposure times. In an ideal, perfect world, we would want to take a single exposure for the full duration of the image (eg a single 10hr exposure). However, should there be a gust of wind, plane trail or satellite trail, or brief cloud cover, the image would be ruined. Also, pixels may become fully saturated (like with stars) and you lose all colour information. Our answer to this is to take shorter subframes and stack together in order to mitigate this risk. But why wouldn't we take subframe of 1s then? Assuming the same camera gain settings, the sensor produces the same amount of "read noise" each time you take a sub. In order to swamp out this read noise, you need to capture a sub for long enough to generate enough signal to overcome the noise. For older CCD cameras where the read noise was quite high, it wasn't uncommon to see subframe exposure times in the order of 10 minutes. However, with these newer CMOS sensors like the IMX571, the read noise is so low that long exposures are no longer necessary. With the aforementioned sensor in mind, using the most popular gain setting (which switches from LCG to HCG mode), you can take an exposure of 5 minutes which can overcome the read noise and without saturating the pixels (in the majority of cases). Other than reducing the number of images you take (resulting in shorter stacking time and storage size), there is not much benefit in taking longer exposures once you've passed the swamp factor for read noise. There may be a case for extended exposure times beyond 5 minutes with regards to very faint objects (like the OIII rich squid nebula), but I have no experience in this area so I'm sure someone else can chip in! |
2.97
#...
·
·
2
likes
|
|---|
|
Sub exposure time will depend on f-ratio, pixel size, sky background flux, filter bandwidth, QE, transmission efficiency, read noise at the chosen gain, and one's "duty cycle" (time between exposures), if the window of opportunity is limited. Anyone who just uses a fixed time without regard to these things is likely not getting the most out of their efforts. There's a big difference in the sub time that's optimal for an L filter vs an H-a filter because the latter lets in far less background flux and the QE is lower in that part of the spectrum for most sensors. Same for Bortle 1 vs Bortle 9 etc. If you're not doing HDR techniques, you want the shortest sub-exposure time that makes you shot noise-limited (which usually means sky-limited). This is also known as "swamping the read noise", which is the one non-time-dependent noise source. Anything longer just wastes dynamic range without having any practical improvement on the eventual SNR of your stack. What the should mean is that all of your different filters would have different optimal sub-exposure times. Where you see a lot of people do 300s across the board is with narrowband filters and that might be for a few reasons: general copycat behavior without giving it much thought or simply because the minimum optimal time for all of them is well in excess of 300s but, for other reasons (e.g. guiding performance), they go for as long as they feel comfortable. When I see people shoot L for the same sub time as R/G/B, that's just a mistake. L accumulates signal and swamps read noise ~3x faster, meaning that they are hurting the DR of their L data unnecessarily relative to their RGB data, unless they are doing so because they're going to bin their color data such that it has the SNR of longer exposures (which itself is an old-school method that is debatable). It usually also a mistake to shoot R for the same sub-time as G/B, for this reason and in terms of matching star brightness profiles in the different channels. See: https://smallstarspot.com/shiny/StackSNR/ |
|
2.97
#...
·
·
2
likes
|
|---|
Richard: Not really if we're talking processed pretty pictures and not scientific data. What you said about swamping read noise still applies. Limiting magnitude is a function of total exposure time absent the practical impact of read noise, not sub time. That said, there's always some residual read noise impact in reality, so always some sense in which it's true, but that's out to decimal places of the limiting magnitude. Any boost is likely far below the level of what will survive processing. Of course a longer sub exposure will show more faint data because, in isolation, it's longer total time. But for stacks of equal time, both with shot noise-limited subs, I doubt one could notice more faint detail in the one with the longer subs. |
7.67
#...
·
·
2
likes
|
|---|
|
There is no "standard" time. It depends on your camera, your focal ratio, and your filters. Actually, a lot of what you hear out there is mostly a hold over from the CCD and early CMOS days when cameras had low quantum efficiencies and high read noise. The need for long sub exposure times pretty much only applies to narrowband filters these days. Its also not just about noise. Have you ever had problems with star colours being tough to bring out? A possible cause of that could be blowing out the pixel wells by sub-exposures that are too long. A big, big thing that people fail to realize is once you hit a certain exposure time there is no benefit to going over it. Longer sub-exposure times do not always mean better final stacks. For example, here are the exposure times I have calculated for my FLT91:  Compare that to my RASA 8:  You can see they are vastly different, mostly due to the focal ratio. If you are used to CCD cameras, a lot of these exposure times will look insane and not long enough, but cameras have come a long way. If you have an hour to kill, I'd suggest checking out this video: https://www.youtube.com/watch?v=3RH93UvP358 Steve Mandel: To actually answer your question though, it seems like a standard time because for the longest time people were told longer subs = better, but 5 minutes was a decent compromise. Again, this is from the CCD days and no longer valid advice with newer CMOS cameras. |
