CCD v CMOS

[ONIKKINEN]

Why does it matter what a single sub looks like stretched or not? If the photons are there they will be separable from noise when stacked.

[Stuart1971]

2 minutes ago, iantaylor2uk said:

What acquisition software are you using? I tend to use the ASI Air Pro, and the images you see on the screen are automatically stretched, so the object would have to be very faint not to show up on the screen.

The image shows up in NINA, each time one is taken and I can see the DSO there, albeit faintly, but as soon as stacked with a small stretch I just get a picture with a green tinge and loads of stars…it has to be really stretched to bring out the image, and with that comes all the noise….

‘My issue really is finding the correct gain setting to use with correct sub length, so as not to maximise full well and loose to much dynamic range…..easy…..hmmmmm absolutely not, this is my issue…..so go short subs and very big gain and hope the stars are not saturated, or longer subs lower gain, which means more noise, but more signal……🤯🤯

[Stuart1971]

2 minutes ago, ONIKKINEN said:

Why does it matter what a single sub looks like stretched or not? If the photons are there they will be separable from noise when stacked.

This is my point, they are not, I get an image full of stars and nothing else until it’s mega stretched…..which brings out all the noise, I would expect after stacking even with very little stretch so see the DSO….but no….🤔🤔

[iantaylor2uk]

If you put the gain up, you will have more signal and lower noise, so that should help

[ONIKKINEN]

2 minutes ago, Stuart1971 said:

This is my point, they are not, I get an image full of stars and nothing else until it’s mega stretched…..which brings out all the noise, I would expect after stacking even with very little stretch so see the DSO….but no….🤔🤔

If you want the image to need less stretching, then maybe up the gain? But not sure i get the problem still. You can have as low as 40 electrons of median signal per pixel and still get a decent image when enough of them are stacked together. Speaking from experience on that one, my stacked images usually have a median ADU value of 120-150. My images still contain noise because they do not have enough integration, but the solution for that seems self explanatory = i just need more integration. By the way, did you image on a mono camera before? I would imagine even a modest mono CCD would steal beat modern CMOS OSC cameras in terms of noise and speed of capture.

[Stuart1971]

Just now, iantaylor2uk said:

If you put the gain up, you will have more signal and lower noise, so that should help

Yes, but what about the negatives of that, lower full well and reduced dynamic range, this is what confuses me with these IMX571 cameras….

[Stuart1971]

6 minutes ago, ONIKKINEN said:

If you want the image to need less stretching, then maybe up the gain? But not sure i get the problem still. You can have as low as 40 electrons of median signal per pixel and still get a decent image when enough of them are stacked together. Speaking from experience on that one, my stacked images usually have a median ADU value of 120-150. My images still contain noise because they do not have enough integration, but the solution for that seems self explanatory = i just need more integration. By the way, did you image on a mono camera before? I would imagine even a modest mono CCD would steal beat modern CMOS OSC cameras in terms of noise and speed of capture.

Yes I had both mono and OSC

i guess I am not explaining very well, with this QHY268c I am trying to figure out an optimal gain setting v sun length, so as not to loose too much full well and dynamic range, up to now I have used the Low gain photographic mode on gain 0 which was stupid as the noise level was at its highest, then I tried the Higher gain mode with gain 0, which I believe is unity, and it was better but still noisy for 3.5 hours of 5 min subs, so am thinking maybe I should up it the 56 gain in this mode, but then I look at the graph for the camera and think am I losing too much full well and dynamic range…..or am I just over thinking it all…..

QHY268C GRAPH showing all 4 modes

 

 

Edited January 17, 2022 by Stuart1971

[iantaylor2uk]

Each sub has a lower dynamic range if you use higher gain, but you choose the sub exposure time so you don't fill the well, and you will get the dynamic range back when you stack lots of frames

[Stuart1971]

 And the full well and dynamic range graph

 

[Stuart1971]

1 minute ago, iantaylor2uk said:

Each sub has a lower dynamic range if you use higher gain, but you choose the sub exposure time so you don't fill the well, and you will get the dynamic range back when you stack lots of frames

Ok, so just really need to make sur I don’t saturate stars … then I k ow full well capacity has not been reached….yes…do I see this in the sub image statistics…?

[Stuart1971]

Looking at the graphs above, mode 1 gain 60 seems the best to me…rather than unity ...but I may well be reading it wrong

[iantaylor2uk]

You should be able to see the histogram in the acquisition software, and you should  be able to use the preview function before you start taking hundreds of images, to check it looks ok

[ONIKKINEN]

9 minutes ago, Stuart1971 said:

Yes I had both mono and OSC

i guess I am not explaining very well, with this QHY268c I am trying to figure out an optimal gain setting v sun length, so as not to loose too much full well and dynamic range, up to now I have used the Low gain photographic mode on gain 0 which was stupid as the noise level was at its highest, then I tried the Higher gain mode with gain 0, which I believe is unity, and it was better but still noisy for 3.5 hours of 5 min subs, so am thinking maybe I should up it the 56 gain in this mode, but then I look at the graph for the camera and think am I losing too much full well and dynamic range…..or am I just over thinking it all…..

QHY268C GRAPH showing all 4 modes

 

 

You are probably overthinking it. At 5 minutes your read noise really doesn't matter anymore, although i would still choose to have a lower one (and shorter subs). If you start saturating stars too much the solution seems simple = shorter subs so they are not saturated. Difficult to say why you have more noise than you think you should, but outside a direct comparison between CCD X and CMOS Y this will be difficult to tell.

1 minute ago, iantaylor2uk said:

You should be able to see the histogram in the acquisition software, and you should  be able to use the preview function before you start taking hundreds of images, to check it looks ok

Looking at the histogram is not that helpful for capture IMO. As long as its not clipped its OK. And black clipping is dealt with an appropriate offset value so that even bias frames cannot contain 0-value pixels. White clipping is dealt with lowering gain or shortening the exposure. If there is signal, it doesn't matter where in the histogram it is if its not clipped.

[centroid]

If nothing else, this has certainly proved to be a very popular subject, as of now some 652 Views, and 61 replies.

At least  is something to pass the time reading, while we sit under the first full moon of the year. 😉

 I just went outside, and its high enough, and bright enough to read a book by. 

[teoria_del_big_bang]

10 minutes ago, Stuart1971 said:

This is my point, they are not, I get an image full of stars and nothing else until it’s mega stretched…..which brings out all the noise, I would expect after stacking even with very little stretch so see the DSO….but no….🤔🤔

I too am still a little perplexed at the issues you are having with this camera, and the frustration it is causing.
I know mine is the mono version so not easy to make a direct comparison but if you want to send ,me some of the subs to compare them to some I have taken you are welcome, I may not see anything significant, but if I can help at all.

I still intend to try gain 0 on mine next clear night and CO does say Wednesday is and with the full moon it will be NB so I can take some at gain 0 and my usual gain 56 so see if there is a noticeable difference.
But it will be with significantly longer suns I know that with my ultra NB filters I get nothing worthwhile on most targets below 5 mins and usually go to at least 10 mins, but as we said before because the nature of your NB filter is different then you should not need longer than 5 min subs.

I guess there could be an issue with the camera itself, very difficult for you to determine this I guess.

Steve
 

[Lee_P]

2 hours ago, Stuart1971 said:

Hmmm, I am not sure about all this, as now the latest CMOS cameras are 16 bit, I was under that impression that longer subs would get better results from these, I have the QHY268c and tried 2 min subs, and got barely anything at all, just a boat load of noise….then as above I shot 5 min subs and still a boat load of noise, and this is with unity gain, and bortle 6 skies, and using an Idas P2 LP filter….at f5.3 with a tak FSQ85…

So I am confused by all these different opinions….on long or short subs….🤯

I use an ASI2600MC Pro -- same sensor as your camera I believe. I always shoot 120s subs, regardless of target or whether I'm using an L-eXtreme or no filter at all. Always 120s. Bortle 8 skies, to make things even harder. Check my gallery and you'll see it can work well

 

[teoria_del_big_bang]

5 minutes ago, Lee_P said:

I use an ASI2600MC Pro -- same sensor as your camera I believe. I always shoot 120s subs, regardless of target or whether I'm using an L-eXtreme or no filter at all. Always 120s. Bortle 8 skies, to make things even harder. Check my gallery and you'll see it can work well

 

Certainly does work well, I know the drivers for these cameras may be different so gains and modes may differ to the QHY version but what gain(s) are you using ?

Steve

[centroid]

I guess the CCD v CMOS contest is pretty much over now, and CCD for amateur astro is now dead , RIP old friend 😢.

I have been looking at a multitude of CMOS images on the forums, and find that while there are some really excellent examples, there are just as many, if not more,  that are way below par, at least to me that is..

It does show that in the right hands, these cameras are capable of returning good results, but how much of this is down to a good understanding of "the beast", and or, the processing skills of the imager.

It is clear to me, that the CMOS camera requires a good understanding of its ability and quirks, far more so than CCD did.

Definitely not 'plug and play' devices, no short cuts, we have to learn to work with them,  and their quirks, to extract the best .

Edited January 17, 2022 by centroid

[Lee_P]

12 minutes ago, teoria_del_big_bang said:

Certainly does work well, I know the drivers for these cameras may be different so gains and modes may differ to the QHY version but what gain(s) are you using ?

Steve

Gain 100, always. Controlled using an ASIAIR Plus. 

[ONIKKINEN]

3 minutes ago, centroid said:

I guess the CCD v CMOS contest is pretty much over now, and CCD for amateur astro is now dead , RIP old friend 😢.

I have been looking at a multitude of CMOS images on the forums, and find while there are some really excellent examples, there are just as many, if not more,  that are way below par, at least to me that is..

It does show that in the right hands, these cameras are capable of returning good results, but how much of this is down to the understanding of "the beast", and or, the processing skills of the imager.

It is clear to me, that the CMOS camera requires a good understanding of its ability and quirks, far more so than CCD did.

Definitely not 'plug and play' devices, no short cuts, we have to learn to work with them,  and their quirks, to extract the best from them.

My camera has been very much plug and play (IMX571 from touptek/risingcam) since i got it. No issues with calibration frames, no amp glow, no fuss about settings. Default offset, default gain in high gain mode and thats about it. Any exposure length over 10s will swamp read noise completely in my typical conditions so why sweat it more than that?

More of these less than ideal results from CMOS cameras compared to CCD cameras could be due to much lower costs involved. The CCD cameras i have looked at for sale have been all extremely expensive, far beyond the typical "just getting into this bit by bit" type of imager. But many of these CMOS cameras are comparable in price to an overpriced smart phone, and not that much of a stretch for many i believe so it could be that there are a lot more beginners out there compared to CCD days.

[teoria_del_big_bang]

1 hour ago, Stuart1971 said:

Looking at the graphs above, mode 1 gain 60 seems the best to me…rather than unity ...but I may well be reading it wrong

In the manual for this camera it does state the following:-

Steve

[centroid]

12 minutes ago, ONIKKINEN said:

My camera has been very much plug and play (IMX571 from touptek/risingcam) since i got it. No issues with calibration frames, no amp glow, no fuss about settings. Default offset, default gain in high gain mode and thats about it. Any exposure length over 10s will swamp read noise completely in my typical conditions so why sweat it more than that?

More of these less than ideal results from CMOS cameras compared to CCD cameras could be due to much lower costs involved. The CCD cameras i have looked at for sale have been all extremely expensive, far beyond the typical "just getting into this bit by bit" type of imager. But many of these CMOS cameras are comparable in price to an overpriced smart phone, and not that much of a stretch for many i believe so it could be that there are a lot more beginners out there compared to CCD days.

An interesting 'take' on the subject, and lot of which is probably very true.

I had no understanding of CMOS sensors used in amateur quality astro cams. The choice was wide, too wide actually.  My decision was based on sensor, and pixel size, plus cooling.

It now seems there are far better choices than the 294 sensor, which I am told is quite old tech now. We live and learn. 🙂

 

 

[MartinB]

4 hours ago, iantaylor2uk said:

The length of time you can go with your subs with a CMOS camera depends on how dark your skies are - if you have very dark skies, then 5 or 10 mins are fine. For most of us with light polluted skies, you are better off with shorter subs, since you are picking up light pollution from the sky background. Also the other rule of thumb is that the narrowband filters usually need 3x the length of subs compared to standard UV/IR filters. So if you're doing 5 mins with a narrowband filter, you really shouldn't be doing any more than 90 secs with just a UV/IR filter. There is a very good video (from Robin Glover) on YouTube and there are Excel spreadsheets around in which you can input your camera, your settings, and sky background, and it tells you the optimum length subs to use. I find stacking a few hundred subs is quite quick in DSS - although it appears other software is not as quick.

Obviously you can use the preview screen to check you are not overexposed with the sub length, gain settings etc that you use.  

The other benefit of using shorter subs is there are less that have satellites flying through so you have fewer to throw away, and some people with lower level mounts may prefer shorter subs to avoid guiding issues. 

Robin Glover explains that there is minimum exposure time required to neutralise the effect of read noise.  He also points out that there is no benefit in increasing sub exposures beyond this for achieving a desired ultimate SNR of the final stacked image.  I have no argument with this at all.  When I got my QSI 532 CCD back in 2010 I used the CCDware sub exposure calculator to work out my minimum sub exposure, it was invaluable.  However, Robin Glover isn't saying that longer exposures are detrimental.  Craig Stark has done a lot of work with CCD cameras examining exposure times and SNR in the real world.   SNR continues to increase beyond minimum optimal sub exposure time.  The limiting factor is the point where pixel response becomes significantly non linear of brighter parts of the image such as bright stars are becoming saturated.  Yes, gusts of wind and rogue clouds give a bigger hit when they ruin a longer sub.  Satellites less of an issue for me since I rarely have less than 30 subs per channel.  One other issue with short subs is time lost through dithering.  You certainly need to be avoiding dithering after every sub otherwise the loss of exposure time becomes substantial.  

 

[Stuart1971]

1 hour ago, teoria_del_big_bang said:

In the manual for this camera it does state the following:-

Steve

So excuse my ignorance, but does that mean upping the gain will help, with shorter exposures and so using mode 1 and gain 56 would be a good idea….?

You have told me before that you use these settings to good effect, so will give them a try, I tried mode 1 gain 0 last time out with 5 min subs, but still noisy, but I guess that maybe somewhat down to the almost full moon….🤔🤔

Edited January 17, 2022 by Stuart1971

[teoria_del_big_bang]

8 minutes ago, MartinB said:

Robin Glover explains that there is minimum exposure time required to neutralise the effect of read noise.  He also points out that there is no benefit in increasing sub exposures beyond this for achieving a desired ultimate SNR of the final stacked image.  I have no argument with this at all.  When I got my QSI 532 CCD back in 2010 I used the CCDware sub exposure calculator to work out my minimum sub exposure, it was invaluable.  However, Robin Glover isn't saying that longer exposures are detrimental.  Craig Stark has done a lot of work with CCD cameras examining exposure times and SNR in the real world.   SNR continues to increase beyond minimum optimal sub exposure time.  The limiting factor is the point where pixel response becomes significantly non linear of brighter parts of the image such as bright stars are becoming saturated.  Yes, gusts of wind and rogue clouds give a bigger hit when they ruin a longer sub.  Satellites less of an issue for me since I rarely have less than 30 subs per channel.  One other issue with short subs is time lost through dithering.  You certainly need to be avoiding dithering after every sub otherwise the loss of exposure time becomes substantial.  

 

I have seen this video sometime ago now but think I need to revisit it and work out some optimal sub lengths.
Due to lack of actual imaging time due to weather last year I did seem to come to a conclusion that with very narrow band filters I needed at least around 5 minutes and generally just put it up to 10 minutes and just left it at that as it seemed to work.
Because of LP I tend not to do LRGB unless the moon is almost non existent and seeing is good, so not so often but have just stuck with 5 minutes for these as well.

Now possibly because of the bandwidth of the ND filters I do need 5 minutes, but possibly not 10 ?? and from what has been mentioned on this thread, and others, I should be far less than 5 minutes on the non NB stuff.

Steve