0.00
#...
·
1
like
|
|---|
|
I was wondering about short subs vs long subs, advantages-disadvantages. Of course short subs require no guiding, and with little movement during the exposure, stars are usually very round. With excellent guiding round stars are present as well, but there is a cost to having excellent guiding. What are other advantages or disadvantages? in other words, what can long subs provide that shorter ones cannot, given the same total time of acqusition? Is color not gathered as well? Thanks for any input!! |
 #...
·
4
likes
|
|---|
|
Good questions! I did not many short sub exposures, and when, then I targeted usually a pretty bright object, like M42. My experience: I cannot get as many details showing up in short exposures compared to long ones, which I think is reasonable. Yes, you do not need excellent guiding, but the image itself does not give as much as one can get with longer exposures. There is also no dithering in place (usually), I can see this as a problem as well. And one shall not forget: processing time! Depending on the sensor of course the processing of 1000 images might take hours, probably days if you use for example the ASI6200. And my computer probably will melt long before a final image comes out. Not saying that short exposures should not be done, but in general I am leaning to 5-10 minute exposures. From what I can tell they produce cleaner images and more detail. There are theoretical works out there that say more than 2 minutes is not needed with our common equipment. From experience I cannot follow this, yes, I understand the mathematical background, but nothing tells you more than trying it. Hence I encourage you to try, target a brighter object and take short exposures, and see what it reveals. Then target a faint object, and you will probably throw those images away since they do not see anything. I have seen some people here on Astrobin who did a marvellous job though with 1s exposures, I do not know how they did this, hopefully someone who did reads this and can shed some light on it. |
|
0.00
#...
·
1
like
|
|---|
|
Thanks for your reply Uwe! Since I use ASILive, the subs are stacked live, as the images are being taken. They are all saved, stacked, in a .fit file. So post processing begins with the fit file and no further stacking needed. The following link is to the 10.2 mag ngc3628 galaxy using 1300 5sec subs, with a little processing. https://www.astrobin.com/6qj5xq/B/?nc=user This is another image of the same galaxy with 1552 5sec subs. I have a lot to learn about post processing, so please excuse... https://www.astrobin.com/dd3747/?nc=user So, were you thinking about another more faint object to test on? Thanks again for sharing your thoughts. |
6.65
#...
·
5
likes
|
|---|
|
Short exposure advantages.-No guiding-Less sensitive to having a good mount (vibrations), winds and seeing , to enable good stars-Low rejection rate of subs (clouds, airplanes)--more subs are good. Can also do a median rejection (or other rejection method) rather than a straight average when combining frames, faint satellite trails can be eliminated without throwing out affected frames. Short exposure disadvantages- There is some delay between subs typically depending on camera, cable, computer--can be 1 to 4 seconds. This is lost time for getting exposure time.- Requires more disk space and longer processing calibrating/registering/stacking. I do lots of short exposures and do not have issues with this. Disk space is cheap (compared to getting what you need for long exposures --great mount, guide camera/scope, dome), and you can just let your stacking runs go on your cloudy nights, and your stack will be ready in the AM. Finally signal/noiseYour camera converts photons to electrons --the ADU (analog-to-digital units, if your camera takes 16-bit images the maximum signal is 65,536 ADU). Your camera sensor doesn't know how you collect the photons. The final ADU signal you get it is exactly the same, no matter how you break it into subs. Signal ADU only depends on the total time of exposure (and the brightness of your target). So, why take longer exposures at all? The only reason is read noise. Every time you take an image (a sub), the camera will add read noise. This depends on your gain--for many cameras higher gain gives less read noise. Obviously then, the fewer subs, the less total read noise. So how long should a sub exposure be to avoid the effect of read noise? There is a good way to decide. The key point is that if your background sky is bright, most of your noise (shot noise) comes from the sky. It overwhelms read noise. So if you sky is bright the read noise rapidly becomes insignificant compared to the sky shot noise. If your sky is very dark, then read noise is a more important contribution. There are some slightly different takes on the optimal sub exposure time. Here is one estimate for your optimal time for your sub, Tsub, considering your sky noise: Tsub = 10 (R^2/sky) Sky=(ADUsky-ADUdark)/Ts - R is read noise in ADU. Measure the standard deviation in the background ADU from a single bias frame. Or use a literature value in electrons for R and divide by the gain to get R in ADU. Same ISO or Gain as your subs will be.-Sky is the sky signal in ADU for an arbitrary sub of length Ts. Doesn't matter how long, as long as your sky gives you a reasonable signal in ADU and is not saturated. You should subtract the ADU for a dark, though generally this is pretty small. Divide by the length of the sub exposure, Ts. Now you have all you need to calculate your optimal sub exposure Tsub. I am in a big city, nearly full moon bright, operating at F6.3--so my optimal exposure is <10 s, more doesn't buy me anything. The darker the site, the longer your subs need to be to be optimal. But remember, in the end the same total exposure will give you the same number of photons, and averaging the subs will improve your S/N, no matter what your sub length was. And in the end, enough of a longer total exposure with shorter subs, could still match a shorter exposure with longer subs. Sorry, this is a bit long-winded. If you want a simple answer, if your sky background is close to at least one-half your full ADU you are doing well with your sub. But for short exposures make sure you set your gain to try as much as possible your read noise. That gain or ISO varies a lot from camera to camera. Since I do everything from the big city I only use short exposures. So far I have not had an issue imaging what I want to image, though I do need more exposure than a dark site. Which is only fair, something is wrong if you can't do more from a dark site. You can see what I have done, both on bright and dark images here @rveregin The key is lots of total exposure--other things are secondary. Clear skies Rick |
|
0.00
#... |
|---|
|
Thanks Rick, for the detailed response! And great images! In the midst of what you wrote, the note about matching the gain to the read noise seemed to be a key. Not sure how I do that. It seems the the gain should be inversely proportional to the time of the sub. The lower the sub time, the greater the gain. On the asi294mc, with a C11, I use 390 for 5 sec subs, which seems to work. Lower, and it is difficult to pick up much. Bottom line also is total integration time. Thanks again for the analysis! |
|
6.65
#...
·
4
likes
|
|---|
|
Higher gain gives you more signal and generally less read noise in many camera's including yours. The only issue with higher gain is that you reduce the dynamic range, so you risk blowing out and saturating bright stars and bright parts of the image. The theory says for S/N you want to maximize the gain and minimize read noise both. For your camera I see gain=350 minimizes read noise. https://astronomy-imaging-camera.com/product/asi294mc-pro-color Gain 390 might be a just a little higher read noise than gain 350, but of course gives you a bit more signal. Probably a wash between these settings. So you are in a good place, again as long as your bright parts are not blown out. And yes, if you use shorter subs you should in general run at higher gain. And if you go longer from 5s to 5 minute or 50 minute subs, you would need to reduce the gain, otherwise you will totally blow out your highlights. Note, you will see some say that any lower dynamic range, which you get at higher gain, is unacceptable. Lower dynamic range is not necessarily bad. Unless you are using the full dynamic range--your highlights are nearly to your max ADU, any extra unused dynamic range is just empty, wasted space that could have been used for your carefully gathered photons by pushing up the gain. And yes, I found longer total integration time not only reduces S/N, it makes bringing up detail easier--you can do more deconvolution and sharpening. It makes colors easier to bring out and to be more brilliant. Rick |
6.89
#...
·
5
likes
|
|---|
|
Hi, With short subs your main concern is read noise. It is roughly the same level or even greater than your signal. Because it is not constant but has a probabilistic level, you can reduce it by taking more exposures. For N exposures, your signal to noise ratio (SNR) is improved by the square root of N. The thing is with short exposures the initial SNR can be very very low. A 60 second sub has 60 times the signal than a 1 second sub, while the read noise is the same. So the 60 second sub has already 60X the SNR than the 1 second sub. You need 3600 1 second subs to get the same result as far as read noise is concerned. The "dead time" spent downloading images between exposures becomes significant in that case: if you have an overhead of half a second to download and write the sub, then the 3600x1second subs will need 1.5 hours, not 1 hour. But even if you ignore this, the fact remains that that it took you 1 hour to reach the same SNR as with a 1 minute sub. If you raise your exposure to 5 seconds, then the initial SNR difference is 12X and you need only 144 subs to reach the same SNR. So you are looking at about 1200 seconds, 20 minutes. If you raise your exposure to 10 seconds you need only 36 subs or 6 minutes total exposure. Not bad at all. The above discussion assumes you used the same gain for the long exposure and the short exposure. In practice you will use a smaller gain for the longer exposure and a larger gain for the short exposure. And on CMOS sensors gain tends to have a direct effect on read noise. For my camera, the ASI178, read noise is 2.2 electrons at zero gain but only 1.3 electrons at gain 300. If you are using gain=0 for the 60 second exposure and gain=300 for the 1 second exposure, then the initial SNR "advantage" is only 35X (instead of 60X) and you need 1225 subs or 30 minutes to achieve the same SNR as the single 60 second sub. For the 10 second exposure, the required number of subs is 12, or just two minutes total integration time. This is not the end of the discussion though. Just because an hour's worth of 1 second subs has the same SNR as a single minute sub, doesn't mean it has the same signal. No sir. An hour's worth of 1 second subs has an hour's worth of signal. Similarly, 60x1second subs has the same signal as a single 60 second sub. If you don't mind the image being a little noisier, you could be very content with the result from multiple very short subs. That's why EAA is a thing, after all. Also, read noise is not really everything. You have to consider detail, dynamic range, star shape. And from a practical point of view you have to consider percentage of keepers for 3600x1second subs (very close to 100%) vs 60x60second subs. You have to consider that for a sub taken at second 00:00:00 and a sub taken at second 00:00:59 the light pollution gradients partially cancel each other when averaging, whereas they work additively on a sub that lasts from 00:00:00 to 00:00:59. It is difficult to think of a dimension where longer exposure wins. Except for SNR. Putting it all together, the only real downside of short subs is the processing and storage overhead. Since a computer capable of storing 1 million subs and integrating 10,000 subs in one go costs about $700 whereas a high end mount starts at about twice that, to me it is a no brainer. At least for broadband and with a CMOS. Narrowband or with a CCD is a very different story. Cheers, Dimitris |
|
0.00
#... |
|---|
|
Thank you Rick and Dimitris for the excellent discussion and insights you have provided! Another area of discussion might be the length of time of the exposure on an unguided scope and any drift which would affect the resolution/detail acquired. Would a decent scope and a 5 sec sub produce better detail than a guided image over a 2 to 5 min exposure? I am sure the accuracy of the guide and quality of stacking and mount drift come into play.. |
12.82
#...
·
3
likes
|
|---|
|
Hi all. A short message because I am on my phone for 3 days. The avantages linked to guiding and tracking have been discussed. But IMO the main goal, specifically on bright objects, is to increase resolution and fight against turbulence. Like we did in the 90's in planetary when ccd became afordable. The goal is to make as short as possible exposition and to sort the frames keeping the best of them. For example, no way one can get such a resolution using long exposures from backyard : https://astrob.in/396710/C/ Cheers, and sorry for my english. JF |
|
0.00
#...
·
1
like
|
|---|
|
Thanks Jeffbax! Very impressive image! And 7000 1sec subs. |
|
12.82
#...
·
1
like
|
|---|
Thanks Jeffbax! Thank you. Actually it is 35000 x 200ms as written at the bottom of the image. Astrobin does not support exposition time under 1 second in its dso description. 1 second is the frontier to start fighting against turbulence. At 500 ms it becomes obvious. CS JF |
|
0.00
#...
·
1
like
|
|---|
|
Thanks for the correction. It makes sense, the method your using. What was the gain set to? |
|
12.82
#... |
|---|
|
In this case it was 500/600. The higher the gain is, the lower is the read noise. Jf |
0.00
#...
·
1
like
|
|---|
|
Hi there, Couldn't agree more with Jeffbax. In my opinion, the primary goal of short exposures is to get a better resolution. Even with 2s subs, i obtained excellent details on some targets like ngc7331 or stephen's quintet ( see in my gallery ) that in my opinion i couldn't have obtained with long exposures and guiding on ground level. The shorter the sub is, the more details you get ; but there is a tradeoff on sensitivity. However, with a well optimized setup, you can use short subs ( around 1s ) and still produce images with signal/noise ratio on par with long exposure images. You just need a good mono sensor witn low readout noise ( imx290 is still the best for now ) and a low F/D ratio. Romain |
|
12.82
#...
·
1
like
|
|---|
|
Cool Romain, je me sens moins seul 😉👍. Merci. Top ton nouveau setup. JF |
0.00
#...
·
·
2
likes
|
|---|
Astrobin does not support exposition time under 1 second in its dso description. Could we ask a modification of such a feature ? |
|
0.00
#...
·
|
|---|
|
what is the cross over point in terms of sub length where the extra detail for DSO starts to set in ? I have been shooting DSOs very successfully at 30 seconds with a DSLR the last 12 months, but recently upgraded to a ZWO-2600MC and I increased the exposure time from 30 seconds to 60 seconds. I notice that many resulting stacked images, seem less sharp,and thought that might be because of the increase in exposure length. I am investigating back focus as another potential source of the loss of detail. very interested in this thread. |
|
12.82
#...
·
·
1
like
|
|---|
David Russell: Hello, well I think short exposures and sorting the frames really increase resolution under 1 second exposure time. In your case it might be a tracking or backfocus issue 😉 CS JF |
|
6.65
#...
·
·
1
like
|
|---|
David Russell: Are you guiding or not? If you are not guiding, then the longer your exposures the more the image will get degraded, unless you have a really good tracking mount and really good polar alignment. Also, I do find if seeing is not great the longer the exposure the worse the image, even as short as 30 to 60 seconds. Though there is even a bigger improvement in IQ with < 1 s due to reduction in seeing variations. Also, if there is any wind or other vibrations, the image degrades more the longer the exposure. One nice thing about shorter exposures and averaging (I use kappa-sigma rejection) is that some of these variations can be averaged out. With long exposures the variations are baked in. If you are not guiding, one trick to see how good your tracking is, is to take a bunch of say 10 second exposures one after the other for maybe 20 minutes. I then stack them in DeepSkyStacker and copy the dx and dy values vs, the reference frame, from the file list in dSS. I paste them into Excel and do a plot. Ideally you want tsee how long it takes for dx and dy to shift less than about 1 to a maximum of 2 pixels. I find that for my mount, which is a good one, I can do 60 seconds unguided (corresponds to 0.5 to 1", well below my seeing which is rarely better than 2.5"). Note there may be bumps due to wind/vibrations, what you are looking for is the general drift due to tracking. If you are guiding and the rms is low compared to seeing, then longer exposures should not be worse due to guiding. If your guiding is poor then shorter will be better. Note guiding cannot compensate for seeing effects, so guiding or no, unless seeing is really good, seeing will have more impact on longer exposures. I would not take subexposures longer than necessary to make read noise small compared to sky background noise. See my discussion above for details on how to do this. I unfortunately am in Bortle 8. So broadband with my low noise CMOS camera, read noise is insignificant for anything longer than 4s subs! With a narrowband filter I need 60 second subs to keep read noise insignificant, since the filter stops most of the LP. Also, if the target is very bright with no faint parts, then read noise is not going to be a factor anyway, you can then go very short subs to get the best detail. It is only when you are trying to capture faint nebulosity near the background sky level that you need to start to worry about read noise and exposure time. Rick |
|
0.00
#...
·
·
1
like
|
|---|
Rick Veregin:David Russell: Thanks Rick. yes I am guiding using PHD2 and EKOS. lately my guiding RMS has been 0.8 to 1.2 roughly. I have been fussy as hell with PA and using EKOS Polar align. last time out I got to 34 arc seconds, which I believe is considered very good or perhaps even excellent. I think I understand what you are saying with how in longer sub exposures the errors get baked in. I am chasing clarity and detail, so investigating how sub exposure length effects those 2 things. I am shooting from Bortle 1which helps. thanks Dave  |
0.90
#...
·
|
|---|
Thanks for your reply Uwe! What about bad sub if its stacked live? Roger |
|
0.00
#...
·
|
|---|
| Sorry for the late response. I believe a poor sub is not stacked as it is rejected by ASIlive, because of difficulty in aligning. Not sure of what the threshold is, but I can have 5% to as many as 25% rejected in the process. |
