Are darks doing anything for me..

[feilimb]

I have captured just a couple of targets at this stage, but have got into the habit of capturing dark frames on my last few targets.  I really don't know if they are worth the effort, as I just tried a stack of my M51 data (Canon 450D @ ISO800, SW ED80 DS Pro @ F6.3, 18x5m light frames), first image with flats, darks and bias, and second image with flats, bias, but no darks.

The only difference I can see from these highly cropped images, is the hot pixel at upper right of 2nd image which was not removed, but I understand these can be removed in post processing anyway.  I'll add a link to my master dark to see if it looks as one would expect.

I would have hoped that adding darks into a stack would help with the noise reduction, but the general noise seems to be identical to me in both images?  Is this the general expected result or should there be a better noise reduction here?  I should note, there was just about 10 dark frames available before the battery died on the camera.

Note: I know that dither is probably the way to go to help with noise reduction also, but in my setup I can't seem to get dither working at all (Ekos on Raspberry Pi 3 controlling guide camera, imaging camera and mount).

Here is a link to the master dark: https://we.tl/JjQK1mosTW

The settings I used in DeepSkyStacker were:

[alacant]

Hi. On my 700d, dark frames introduce more noise but using a few pixels dither between light frames then stacking with median or Kappa-Sigma without them removes all the hot pixels and a good deal of the noise. HTH.

[ollypenrice]

I see nothing gained via your darks. A hot pixel filter at the stacking stage would zap the one you have. When DSLR imagers like Tony Hallas advise against them I can't say I'm surprised and I don't use them myself even with a set-point cooled CCD where you know the temperatures match. You never know this with an uncooled camera.

I can't help with the dither settings but you really should prioritize getting it to work. It will also fight the very high level of colour noise in your background. ('DSLR colour mottle.') If you can't get it to work then you can easily dither manually once every few shots.

Olly

[pipnina]

1 hour ago, ollypenrice said:

I can't say I'm surprised and I don't use them myself even with a set-point cooled CCD where you know the temperatures match.

Do modern cooled CCD's not produce enough dark current to warrant using darks? Or do you not use exposures long enough to make them noticeable on a cooled camera? Rather interested since my DSLR creates a noticeable level by about 3 minutes.

[ollypenrice]

57 minutes ago, pipnina said:

Do modern cooled CCD's not produce enough dark current to warrant using darks? Or do you not use exposures long enough to make them noticeable on a cooled camera? Rather interested since my DSLR creates a noticeable level by about 3 minutes.

I routinely use 30 minute subs and rarely less than 10 minutes, so I get plenty of noise! Rather than subtract a dark I subtract a master bias and apply a defect map. In my case the defect map is a thirty minute master dark with 2000 ADU clipped off the bottom. I find this both easier and more effective than subtracting darks on our Kodak chips. On my Sony chip I just subtract a master bias and that's it. No need for the defect map. On the Kodaks this does leave a small patch of amp glow in one corner but that's a ten second fix in Photoshop.

Olly

[wxsatuser]

A lot more subs will help.

http://www.samirkharusi.net/sub-exposures.html

[ollypenrice]

3 hours ago, wxsatuser said:

A lot more subs will help.

http://www.samirkharusi.net/sub-exposures.html

By far the best way to reduce noise, for sure.

Olly

[feilimb]

6 hours ago, wxsatuser said:

A lot more subs will help.

http://www.samirkharusi.net/sub-exposures.html

Many thanks for the link, that article looks very useful. Olly, as you mention dithering is probably a priority - I'll continue to seek for support on the bug in the software I am using and if all else fails can try a manual dither approach as you say - maybe there is some way to automate a bump of a few pixels between subs, by turning off tracking for a second and then back on again or something like that.  Maybe the darks would prove more useful if I was taking much longer subs eg. 10 minutes plus, I may try much longer subs on my next session and see how they come out (I have not determined max exposure time yet for my dark sky site which has medium light pollution).

[Adreneline]

15 hours ago, ollypenrice said:

Rather than subtract a dark I subtract a master bias and apply a defect map. In my case the defect map is a thirty minute master dark with 2000 ADU clipped off the bottom.

Hi Olly,

Probably a stupid question but what process/procedure do you use to clip 2000 ADU off the bottom?

Thank you.

Adrian

[ollypenrice]

6 minutes ago, Adreneline said:

Hi Olly,

Probably a stupid question but what process/procedure do you use to clip 2000 ADU off the bottom?

Thank you.

Adrian

Not stupid at all. I stack and calibrate in AstroArt and simply follow their instructions. There's an Arithmetic menu under which you have the option to 'clip,' so you type in 2000 and the result is a hot pixel map, AKA Defect Map. Why subtracting the bias separately and using the defect map works better than 'all in one' darks I don't know, but I find that it does and it's much easier because different sub lengths all use the same calibration files.

(With normal darks I'd sometimes get walking noise on the RGB. That no longer happens and despite the considerable noise of the Kodak chips I frequently have no pixels whatever to clean up in the final image. I can't guarantee that this will work for everyone, though.)

Olly

[Adreneline]

1 hour ago, ollypenrice said:

I can't guarantee that this will work for everyone, though.

I've used the Cosmetic Correction process in PI and as you point out it has the advantage of using a MasterDark; I used the longest Darks I have - 1200s. I have also used the CC process via the Batch PreProcessing Script and used a MasterBias in place of Darks. I think there has been an improvement in the stacked image; it's hard to say in many cases probably because the LP and 'seeing' were rubbish at the time I acquired the data!

Thanks again.

Adrian