What should single dark frames look like?

[Jonny_H]

Hi all,

As above - what should they look like?  The reason I ask is that i have seen a few people post up dark frame examples that they say are stretched and look like a smooth grey colour with no noise whatsoever.  Mine however are a rather different story.  I have attached a FITS example from my latest set which were 600s @ 150 gain and cooled to minus 10.  Camera is a 533MC Pro.

Unstretched the image is totally black however when i apply a stretch, whether it is in PixInsight or via ASIStudio - the image becomes incredibly noisy and tons of hot and cold pixels everywhere.

So what are they supposed to look like?  Like the smooth grey images I have seen or like mine?  If the former.....what am i doing wrong!?

 

Thanks,

Jonny

Dark_600.0s_Bin1_gain150_20210331-205503_-10C_0001.fit

[teoria_del_big_bang]

I think thats pretty normal for a single dark frame that has been auto stretched.

Stack 50 of them together with the correct parameters and it will look a lot better.

If doing an auto stretch on what is effectively a black frame will stretch it a huge amount so will pick u any small discrepencies.
Even on a stacked image an auto stretch will pick something up.
In Pixinsight do an auto stretch on a single frame then use exactly the same stretch (so not an auto stretch) on a correctly stacked master dark and you will see a big difference (well I do anyway if I am doing it right 🙂 )

Steve

Edited April 6, 2021 by teoria_del_big_bang

[The Lazy Astronomer]

I will caveat what I'm about to say by saying I've never actually looked at my darks (naughty, l know!), but I think a dark frame should be as you describe: pretty much completely black with a smattering of hot pixels (and amp glow, if your camera is affected by it). Flat frames should be a nice uniform grey.

[symmetal]

That single dark frame looks normal. The final master dark after stacking should be left with the fixed pattern 'noise' of the camera which includes hot, warm and cold pixels, amp glow and the bias. The random noise visible on the single dark frame will  be reduced significantly by stacking a sufficient number of darks, the more the better, but the improvement in noise reduction by adding an extra dark frame reduces each time, so there comes a point where it's not worth adding more frames.

Alan 

[teoria_del_big_bang]

21 minutes ago, The Lazy Astronomer said:

I think a dark frame should be as you describe: pretty much completely black with a smattering of hot pixels (and amp glow, if your camera is affected by it).

I would still say that this is almost true for an averaged stack of say 20 or more frames but a single dark frame will have random noise in there, especially on a non cooled DSLR, hence why we average a umber of frames and up to a point the more the better (20 usually minimum and 50+ common) to reduce any random noise.
How it looks also depends on the amount it is stretched.
Auto stretched images will rarely be just black which is why when comparing one image to another by stretching them you must make sure they are stretched exactly the same.

Look at THIS article

21 minutes ago, The Lazy Astronomer said:

Flat frames should be a nice uniform grey

Unstretched yes but when stretched it will have blobs and patches of different shades of grey depending on how much dust and dirt are on the optics and other things like vignetting.

Steve

Edited April 6, 2021 by teoria_del_big_bang

[Jonny_H]

Thanks all for your responses.

So perhaps some of the posts I have seen where the 'single' dark image example is a nice uniform dark grey/black - I assume they are referring to the single master dark as opposed to a single dark frame in the literal sense.

[symmetal]

3 hours ago, Jonny_H said:

Thanks all for your responses.

So perhaps some of the posts I have seen where the 'single' dark image example is a nice uniform dark grey/black - I assume they are referring to the single master dark as opposed to a single dark frame in the literal sense.

It still depends on how much the image is stretched before viewing.

Here's my master dark for the ASI1600 comprised of a stack of 64 frames of 90s exposure. It's given a linear stretch as shown by the bottom of the two histograms, so that it's displaying from 0 to 3024 ADU rather than the 0 to 65535 ADU on the top histogram. It's showing the full frame so is zoomed to 17% of full size. Visually it looks uniformly grey.

Here's the centre of the above stretched image, indicated by the box top left, at 100% zoom, so full size. Individual warm and cool pixels are visible which are up to around 500 ADU from the average background ADU of 990 ADU. It still looks uniform unless you're pixel peeping.

Here's the same full frame master dark displayed in Astroart at 16% zoom with a very severe stretch. Amp glow is clearly visible along with severe noise over the whole frame. The white parts of the amp glow are at about 1020 ADU with the average background at 990ADU.

Finally here's the centre of the above image at 100% zoom. A mass of noise though the difference from the black to white pixels is around 20 ADU here.

You wouldn't say the last two images are uniformly grey, but all four images are of the same master dark file. 🙂

Alan

[andrew s]

Looks can be deceptive as @ollypenrice has shown. Look at the histogram. It sould, in an ideal world, show gaussian noise offset from zero by the bias level.

In reality it will have additional noise e.g. telegraph noise in CMOS sensors and a high signal tail to the noise due to hot pixels.

Some structure can show when zoomed in but as long as its static it does not matter.

Regards Andrew 

[symmetal]

7 hours ago, andrew s said:

Some structure can show when zoomed in but as long as its static it does not matter.

Yes, I should have stressed that the 'noise' shown by the highly stretched stacked dark image is predominantly the static 'fixed pattern noise' of the sensor and not just random noise. The fixed pattern noise is very temperature dependent which is why darks need to be temperature matched to the lights.

Alan