Dithering Vs Darks (DSLR) - how many Darks anyway?

[Jonk]

I've just tried to add bias files as darks in DSS - no problem, but then adding the same bias files as bias files, it doesn't work (they won't add).

Just waiting for a stack to complete using bias as darks. I understand the reasons so hopefully, there will be a different result.

Not for me to use use longer term anyway, as I'll continue to use PI as I save for a CCD. I was just curious to see what difference it made to the stack.

[Jonk]

It crashed and I'm tired so I'll try again tomorrow!

[wimvb]

As I understand it, in the PI calibration process, the master bias is subtracted from the dark frames during calibration. The master dark therefore does not contain any bias information.

The reason for doing this is according to the PI website, that dark scaling works better when read noise is absent.

When calibrating the light frames, the master bias and master dark are subtracted from the lights.

If you want to calibrate with bias as darks in PI, you just skip the darks alltogether. Then use cosmetic correction to clean up hot pixels, and a very low pixel rejection parameter during integration.

[Ouroboros]

Wouldn't the calibration process subtract bias from the lights twice if bias weren't first subtracted from the darks - because dark frames inherently include the bias signal?  In fact that makes me wonder why the recommended procedure is to take bias frames at all. 

[Ouroboros]

6 hours ago, wimvb said:

If you want to calibrate with bias as darks in PI, you just skip the darks alltogether. Then use cosmetic correction to clean up hot pixels, and a very low pixel rejection parameter during integration.

Are the hot pixels not present in the bias frames? If they are I would have thought they would be subtracted when bias data is subtracted from the lights. 

[Ouroboros]

Just checking to see myself the effect of subtracting  darks .... I processed the same lights without darks subtracted (first image) and with darks subtracted(second image). both images were stretched the same amount.

I don't know the reason for the "carpet flecking" appearance of the un-dark subtracted image. Those flecks do not appear in the master dark or unprocessed subs. 

Would I be right to assume that dithering would average out the flecking? 

15 lights, 10 darks, 20 bias frames. Small section. Camera 450D un-modded.   

[Stub Mandrel]

8 hours ago, wimvb said:

As I understand it, in the PI calibration process, the master bias is subtracted from the dark frames during calibration. The master dark therefore does not contain any bias information.

The reason for doing this is according to the PI website, that dark scaling works better when read noise is absent.

When calibrating the light frames, the master bias and master dark are subtracted from the lights.

If you want to calibrate with bias as darks in PI, you just skip the darks alltogether. Then use cosmetic correction to clean up hot pixels, and a very low pixel rejection parameter during integration.

My understanding is that DSS does exactly the same.

[wimvb]

Defect pixels should be in any frame. The number of bad/hot pixels (as opposed to completely defect pixels) will increase with temperature and become more obvious with long exposures. They usually don't show up in bias frames. The easiest way to see the difference is to compare a bias frame to a dark frame, exposed long enough.

If you don't use darks, you get rid of bad pixels by using a filter of some sort during image calibration and stacking.

In your comparison, did you process the images with the same settings, just leaving out the darks, or did you use a more aggressive filtering routine?

Dithering will make sure that bad pixels are never in the same place on registered images, making it easier to remove those bad pixels during stacking. Ideally, bad pixels never make it to the averaging stage of stacking.

[Ouroboros]

15 minutes ago, wimvb said:

Defect pixels should be in any frame. The number of bad/hot pixels (as opposed to completely defect pixels) will increase with temperature and become more obvious with long exposures. They usually don't show up in bias frames. The easiest way to see the difference is to compare a bias frame to a dark frame, exposed long enough.

If you don't use darks, you get rid of bad pixels by using a filter of some sort during image calibration and stacking.

In your comparison, did you process the images with the same settings, just leaving out the darks, or did you use a more aggressive filtering routine?

Dithering will make sure that bad pixels are never in the same place on registered images, making it easier to remove those bad pixels during stacking. Ideally, bad pixels never make it to the averaging stage of stacking.

Yes. Same settings, leaving out the darks. 

[wimvb]

Just leaving out darks should give worse results; darks do have a function in AP. If you decide not to use them, you have to replace their function. A bad-pixel filter and aggresive pixel removal during stacking has that same functionality.

In your image (- darks), try cosmetic correction and hot pixel removal as well as lowering the clipping threshold in sigma clipping. This should give better results. If you get the same results as with darks, you can use either method.

Good luck

[Stub Mandrel]

3 hours ago, wimvb said:

In your comparison, did you process the images with the same settings, just leaving out the darks, or did you use a more aggressive filtering routine?

As far as I could.

[sharkmelley]

6 hours ago, Ouroboros said:

Just checking to see myself the effect of subtracting  darks .... I processed the same lights without darks subtracted (first image) and with darks subtracted(second image). both images were stretched the same amount.

I don't know the reason for the "carpet flecking" appearance of the un-dark subtracted image. Those flecks do not appear in the master dark or unprocessed subs. 

Would I be right to assume that dithering would average out the flecking? 

 

The "flecking" is caused by the image drifting slightly from frame to frame.  When the stars in each frame are registered and stacked then in the final image it appears that it is the hot pixels have drifted instead, causing the flecking appearance.  This would be reduced if a sigma rejection stack is used.  The purpose of dithering is to apply a random drift between each frame which makes the drift pattern less obvious to the eye but more importantly allows sigma rejection to work much better.

However there's one other important reason to perform a dark subtraction: dark subtraction prevents the hot & warm pixels spreading out into "blobs".  The image below shows how hot red, green and blue pixels get spread out during the colour filter array demosaicing process (sometimes called debayering) of a colour sensor (i.e. the conversion of the raw data into a full colour image).

In your flecking example each fleck is the result of a hot or warm pixel spreading into a blob and then drifting from frame to frame.  That's why it is so obvious in the image.

In my opinion it is best to treat hot & warm pixels before they get the opportunity to spread out into a blob.  Hot pixels are best dealt with by cosmetic correction and warm pixels by carefully calibrated dark subtraction.

Mark

[bottletopburly]

If I invest in a cooled CCD will this all go away anyway?  Many thanks for any advice.  Here's my dithered Elephants Trunk Neb, over processed. (My search for an alternative to Photoshop continues)

 

Try Startools for processing  free to try, all features except the save option which unlocks on purchase