osc noise confusion

[alacant]

Hi. I'm thinking of a sideways step from dslr to cooled cmos osc for the hotter months. I see a lot of noise and patches in examples and a lot of discussion about dark current noise or whatever which I don't get with my dslr. With the dslr I can lose the dark frames by dithering.

Questions.

1. must I use dark frames with a osc?

2. must I crop to remove the patches?

Here is a dark stretched .fits from the osc in which I'm interested. You can see loadsa noise, banding and the patch top right.

3. Is this removable say with flat frames?

TIA

 

 

 

[vlaiv]

You should distinguish between dark noise and dark current signal.

Most of the noise you see in that image is read noise. With cooled cameras dark current noise is rather small and insignificant component (compared to dark current signal and read noise / signal).

With calibration you don't remove noise (random component) - you remove signal component (non random contribution of both read process and dark current). If you want to capture only light in your images, then removing these signal components that come from sources other than light is important.

You overcome noise by making sure that you have adequate SNR - by using long exposure and/or stacking multiple exposures.

Dithering that you did, help with both read and dark signal noise - irrespective of calibration. You should always do proper calibration to get best / most accurate results. When doing DSLR imaging you can "get away" with not using dark frame subtraction for several reasons:

1. Sensor has very uniform / small dark current signal - so no "hot zones" - or amp glow that you should calibrate out.

2. You don't have set point temperature so reproducing darks is very hard to do

3. Possibly DSLR, because it is aimed at consumer market - daytime photography, has some sort of internal optimization that "masks" away dark current - like there are noise reduction modes in many models. Some even have dark subtraction as option for long exposures.

With cooled OSC sensor - I strongly recommend that you do full calibration (darks, flats and flat darks) - as it will bring best results. Also, astronomical cameras are rather "bare" - all "brain" components, or consumer electronics that comes with cameras has been removed, so it is up to you to process images properly - also no white balance to set, that is something that you also must do manually (or via suitable algorithm).

You can't really judge amount of noise by looking at stretched frame, unless you have frame from different camera, exactly equally stretched to compare it to. Rather compare noise stats on both cameras.

And to answer your questions in order:

1. Must is a strong word but I would definitively suggest that you do so - with set point temperature cam it is really not that hard - you just leave it for couple of hours during cloudy night or even during the day to gather subs. You only need to do it once in a while (like once a year or twice a year if you happen to have large temperature swings during year - so you'll have summer and winter set of darks, at least I do because I can't reach -20C in hottest summer months, so I shoot at -15C).

2. You will not need to crop anything to remove bright spots, provided you do proper calibration with darks (see answer to question 1 ). If you choose not to do darks, then it will depend on image / target and achieved SNR.

3. Flat frames deal with light signal - they correct the fact that not all surface of sensor receives / registers same amount of light signal. They can't be used to correct things that are not related to light signal - like read signal, or dark signal. On the other hand if you use flat frames (and I advise you to do so), you will need to remove other types of signal from flats (by using flat darks) and insure that flats are applied to light signal only (by removing other types of signal from your light frames - or calibrating by darks - see answer to question 1 )

HTH

[ollypenrice]

Lots of people assume that they can asses a camera's noise by looking at its dark frames. I'm not one of them. What interests me is how noisy my calibrated final stack really is, and at this point I find something interesting: my 'noisy' Kodak 11 meg chip gives a stack which is a peach to process. My 'low noise' Sony chip gives results which have had me devising new techniques for getting a decent result out of them. Now I'm happy with the results of the Sony chipped camera but, and this is a big but, it has taken me some time to work out how to do so. For me 'noise' is whatever is getting in the way of the picture. The problem I find with the Sony chipped camera - and no end of guests have been privvy to this post processing battle - is that it seems to be sensitive to very low signal but not to the very-very low signal of the background sky at my dark site. So I get background pixel values, in a normal stretch, ranging from 9 to 22. The noisy old Kodak, calibrated and stretched, gives a range of maybe 20 to 22. I know which I prefer.

You can theorize till you're blue in the face but what works is experimentation. Try stacking with darks, with bias as darks, with bias as darks and bad pixel map, etc etc till you find what works for you.

Olly

[alacant]

Thanks guys. So, with apologies in advance, which do you reckon would show the biggest improvement -if any- over a Canon 700d:

asi1600 or asi294

?

TIA

[ollypenrice]

11 minutes ago, alacant said:

Thanks guys. So, with apologies in advance, which do you reckon would show the biggest improvement -if any- over a Canon 700d:

asi1600 or asi294

?

TIA

Only listen to people who've tried them. I haven't tried either.

Olly

[Allinthehead]

That 294 looks very good on paper. 14 bit and much higher full well. As Olly said listen to the users of each.

I have a 071 osc and a 1600 mono so i can only say that the mono version of the 1600 is good but has its flaws. You get a strange diffraction pattern in combination with some scopes on the brighter stars. You should check if the osc is the same. 

[alacant]

13 minutes ago, Allinthehead said:

I have a 071

Thanks for your reply. Would you say your 071 was better than a dslr? IIRC, it's the same size(?)

Cheers.

[Allinthehead]

8 minutes ago, alacant said:

Thanks for your reply. Would you say your 071 was better than a dslr? IIRC, it's the same size(?)

Cheers.

I used a 600d and the difference is huge. I also had a Sony A7s and the 071 is better for my conditions. 

 

[vlaiv]

I'm a bit skeptical about 294 because it features "HCG mode", or there are two distinct gain regions with different noise characteristics. Otherwise looking at the specs I would be for it, but if you want to exploit full well capacity of ~64K you have to settle with more then 7e noise (very high read noise for CMOS sensor). But if you decide to go above unity gain, then read noise drops below 2e - which is good in CMOS terms. If it were right at the boundary of unity gain, so that unity gain had 2e noise, then I would say, yeah go for it and use unity gain setting. But if you look at the graphs, unity gain is 117, and break point where read noise starts to be below 2e is 120.

All of this might not mean much in terms of usability, but I would not know how to approach it (what would be settings to likely produce good results) off the top of my head, so I'm reluctant to advocate for it.

I have not yet had problems with 1600 and reflections, but there still might be some room for that to appear since I have not tried out all combinations of focal lengths / spacings / scopes and reducers - correctors with it.

For 071 I've only heard good things - a definitive improvement over DSLR.

[alacant]

14 hours ago, vlaiv said:

this might not mean much in terms of usability,

Thanks anyway. It's given me some new vocabulary to come to terms with.

I'll stick it on the camera thread to see if anyone has used one.

Cheers