Simple questions on darks

[MikeP]

Sorry to start a new thread but the existing threads that discuss darks are a bit complicated for me.  I have two questions:

My imaging camera is an Atik 16HR, should I be using dark frames at all?  Other threads seem to imply that if your camera is cooled, it is not necessary.

If I should be taking darks, what I'd read / been told was that darks should be the same duration as the lights and taken at the same temperature.  It seems that this is not the case for PixInsight, so does this mean I can take one set of darks, create a master and use it for everything I ever image in the future?  I can't believe that is correct, but it would make me happy. 

Thank you for any guidance you can provide.

Mike

[Nigel Ball]

Mike

Yes you do need darks - even if your camera is cooled. These should be taken at the same temperature as your lights -20°C seems to be fairly common.

I personally take darks of the same length as my light frames. The first task when I got my CCD camera was to build up a library of dark frames for future use. I suggest you set your camera running during the day. Now it is winter cooling wont be an issue. Take 20 or more at each duration then combine for a Master.

HTH

Nigel

[mmirot]

Making a library of darks and bias frames is fairly standard. It is a good practice to gather a new set every six months to year.

The reason is CCDs do age slightly. This usually identified as a few new hot pixels every year or so. 
It is the unfortunate effect of cosmic rays hitting our sensors.

If you imaging from inside a coal mine your sensor will never change. Just hard to gather regular photons 

Max

[Jack Harvey]

I have read in several places that the Sony chips have so little dark noise that darks are not necessary.  I have never had a camera with a Sony chip so cannot comment personally but perhaps others here can.

[MikeP]

Thank you all for your answers.

Sony chips have so little dark noise that darks are not necessary

There is another thread somewhere relating to a Starlight Xpress cooled camera that implied darks were not necessary - if I find it again, I'll see if I can figure out if it relates to the same chip.

If you imaging from inside a coal mine your sensor will never change. Just hard to gather regular photons

I might as well be in a coal mine the way the weather has been here in Sussex.

Take 20 or more at each duration then combine for a Master.

OK, it is time for a stupid questions again, but better safe than sorry .....  If I exposed 20 darks for 1 minute each, 20 for 2 minutes and so on up to my longest expected exposure, would I then combine 1 minute darks to create a 1 minute Master, then combine the 2 minute darks for a 2 minute Master and so on?  In other words a set of Masters varying by duration - not just one "mega" Master.  Sorry, but I really am this thick.

Mike

[Carlos Milovic]

Just make a master dark for the longest exposure time you made. Also, take lots of bias (>50) and then you may rescale down the dark to the shorter exposures.

[MikeP]

Thank you Carlos - not the answer I expected but the "Master Calibration Frames" tutorial now makes sense.  Hooray.  Unfortunately, I do have two further questions, sorry.

Here is part of a post (from another thread) from Mike Reid that was not answered ....

I have a typical LRGB set of calibration frames where I'm shooting L at 1x1 binning and RGB at 2x2 binning.  I tried to follow Vicent's tutorial and these are the steps that I did to generate my 8 master calibration frames,

  01. integrate 1x1 bias frames                 => MasterBias_1x1.fit
  02. integrate 2x2 bias frames                 => MasterBias_2x2.fit
  03. bias calibrate 1x1 dark frames
  04. bias calibrate 2x2 dark frames
  05. integrate 1x1 calibrated dark frames      => MasterDark_1x1.fit
  06. integrate 2x2 calibrated dark frames      => MasterDark_2x2.fit
  07. bias and dark calibrate L flats frames
  08. bias and dark calibrate R flats frames
  09. bias and dark calibrate G flats frames
  10. bias and dark calibrate B flats frames
  11. integrate calibrated L flats frames       => MasterFlat_L.fit
  12. integrate calibrated R flats frames       => MasterFlat_R.fit
  13. integrate calibrated G flats frames       => MasterFlat_G.fit
  14. integrate calibrated B flats frames       => MasterFlat_B.fit

Does this seem correct?

Is it correct to create masters for bias and darks for each each binning mode?
Are his steps 03 and 04 correct?  I'd assumed that the master dark was just an integration of all the darks.

Mike

[DaveS]

Hi Mike

I always used to religously take 'darks', equal to my light subs, but quite frankly I found that with the Sony 285 CCD in the SXVF-H9C and now the SXVF-H9, and subs up to 10 minutes duration or more, they really weren't needed. Except to deal with any 'hot pixels'.

'hot pixels' are easily dealt with by using 'dithered' guiding, and combining/stacking with SD Mask, which will reject them.

I do however, always take and subtract Bias frames. The Bias is of course part of the dark frame, but just taking the bias on it's own is infinitely quicker.

BTW, if you are now getting into imaging 'proper', you know where you should be!!  

Dave

[Carlos Milovic]

If the binning process is done by software, then probably you should just take 1x1 calibration frames... but, since I don't know if that's true, I would recommend taking both types. It is preferable to do extra work than mess things up

[vicent_peris]

Hi Mike,

think on dark frames as collecting another kind of data, just in the same way as if you were collecting photons. You have a dark signal in your sensor, and you want to collect it with the less possible noise. So take long exposures (just to not saturate your hot pixels), and collect many many of them. Tow point here:

- This is the same work as collecting photons. Your sensor has a dark signal of xxx e-/sec. So the longer time you spend collecting dark signal, the better SNR you will have.

- It's better to do longer exposures than shorter ones. With longer exps the dark signal will be stronger against read noise. So it's better to do 10x20 min darks than 40x5 min ones (assuming your hot pixels don't saturate in 20 min).

OTOH, always do *a lot* of bias. This is very important for the dark rescaling algorithm of IC.

For the moment, I advise you to take your darks at a similar temperature as your light frames (+-5ºC?). We are going to implement an improved version of the rescaling algorithm that will allow us to manage with bigger temp differences.


Regards,
Vicent.

[MikeP]

Thank you Dave, Carlos and Vicent.  I did wonder how temperature difference affected rescaling, I guess the new algorithm will have a parameter to allow the difference to be input.

I'm quite excited about capturing sets of bias, dark and flat frames because I know how to do it, whereas capturing lights requires skills I'm still acquiring and it also requires the weather to cooperate. 

So what is probably a tedious chore to you old hands is real progress for me.  Thank you very much for the help.

Mike

[Carlos Milovic]

The problem with temperature, is that dark noise is not too well fitted by a linear function. So, it is better to stay between a few degrees around the temperature at the capture.

[vicent_peris]

The new algorithm will be completely automatic, too.

Anyway, the only problem you will have now is that, if the temp difference increases, the hot pixels won't be completely corrected. But this is not a serious problem IMO; they can be corrected with a rejection algorithm in the integration step. Think that *all* the pixels have thermal signal, so hot pixels do only a small contribution in terms of noise to the whole image.

Best regards,
V.