Noisy Dark frames?

[Hackleby]

Hi everyone,

I have been struggling for some time now with DSS and have found that single exposures always look better than a series of stacked images. I have persisted but keep coming back to the same conclusion.

I recently purchased "A Guide To Astrophotography" and after watching several of the tutorials decided to have a go at subtracting darks.

What became immediately obvious to me was the difference between My dark frames and those from the tutorial. Granted I'm a beginner but surly the dark frames I took should not contain so much noise. The attached image shows, Left to right, a 600s light frame at 50% and 500%, My 600s dark at 50% and 500%, Tutorial  600s Darks at 50% and 500%. 

Are my dark frames normal or are they noisy and if so what could be contributing to this?

Stacking images in DSS with darks, flats and bias washes the picture something terrible and there doesn't seem to be enough data left for stretching after wards.

Any advise is welcomed.

Thank you.

[Tinker1947]

I never have bothered to learn to much about dark frames or any other sort, but  the ISO used could have a bearing on the amount of noise, when i used 1100D the ISO was set at 800, all Noise reduction in the menu's were set to off....

[Hackleby]

The ISO used in my light and dark frames is 800, the dark frame from the tutorial is at 1600.

Camera used is a Canon 6d although my 1100d has same issues, noise reduction turned off since I started AP as recommended by most people. 

[joecoyle]

Hi,

From the file name it looks like you've applied a standard deviation to the dark frame? Is that a single frame? As far as I know dark frames should be median combined...

Light and dark frames need the same ISO and duration and temperature. A lower ISO should be less noisy not more.

Joe

[Hackleby]

The file names were allocated to each sub by the camera control software "BackYard EOS"

All three subs used in the image are single and unprocessed, other than converted from RAW to TIFF to import into PS to create attached image.

I would also have thought that my lower ISO would have produced a cleaner image.

I have also compared the lights and darks on the tutorial software and can clearly see all of the noise in the dark on the light.

How ever I see none of the noise in my darks on my lights so something must be wrong with my camera setups.

I have followed several walk through's (is throughs a real word) which are basically all the same so I am unsure what the problem is. 

[glowingturnip]

have they all got the same stretch to them ?  If you apply the same stretch that you've used for your light to your dark frame then you would indeed expect to see a predominantly black frame with a few noisy pixels (which should match your light).  If you did a stretch to show *everything* on your dark though, then you will be amplifying tiny differentials between pixels which will show as noise.  Maybe that's what you're seeing above

(and should it be 'walks-through' ??  )

[kalasinman]

AFAIK calibration should be applied in a linear (unstretched) state. Ignore the differences between your darks and the tuatorial. Different cameras at different exposures, ISOs, and ambient temps will show different amounts of noise. Don't worry about noise in your dark frames, worry if the dark frames don't reduce noise in the finished stack.

[FunkyKoval35]

Another thing is that astrophptography with not cooled DSLR during the warm nights(above 0 - 5 *C) is some kind of cross-purposes Noise, noise, noise....

[Hackleby]

No stretch has been applied, just converted from RAW to Tiff to enable import into PS.

I'm still at a loss as to why my lights do not contain the noise on the darks like the tutorial lights do.

Think I need to worry as the darks are not reducing noise, they are reducing the amount of data in the image.

[dph1nm]

just converted from RAW to Tiff to enable import into PS.

This will almost certainly have applied a stretch unless you explicitly told whatever software you used not to. Your dark definitely looks highly stretched. Linear frames nearly always look mostly black (including light frames!).

NigelM

p.s. remember darks will always have read noise in them, irrespective of how much dark current there is.

[Tinker1947]

I loaded a Dark frame into PS its completely black, stretching it by moving the centre slider to the left results in a very noisy red specked frame....

[lensman57]

Sorry for the double post. I have deleted it.

A.G

[lensman57]

I am afraid that your statement that a single frame looks better than a stack is incorrect. If you stacked  the frames and end up with more noise and not enough signal then something is not right. This is usually caused by incorrect  or mismatched calibration frames. I would not bother with the Dark frame first if I were you and just use a large number of Bias and Flat frames to start with and see if the image improves. You should also be merciless with the Lights and examine each frame before putting it in the stack. A single bad Light frame with low signal or misshaped stars will degrade the entire stack. I have attached two small jpegs of the Cocoon nebula taken a few nights ago. One is a single sub and the other is a stack of only 5 subs, these have only had calibration and a stretch with a straight STF parameters applied to them in Pixinsight. I leave it to you to guess which is which. I normally capture at least 10~12 subs and often 16~20 for stacking. I threw away 4 subs from this imaging session since they were ruined by thin passing clouds and had low signal. In theory a single sub will have less noise than  a stack but it will have much lower signal so the stack wins.

Regards,

A.G

[Stub Mandrel]

Your dark is misleading it looks like you have resampled at 500% and 'smeared' the pixels. You need to zoom in past 100% so you can see individual pixels and do a screen grab.

This is what one of my darks (which looks black with isolated dots as a RAW) looks like when saved by RAW Therapee as a tiff and zoomed in. Without zooming it looks just like yours in RAW Therapee (note that RAW Therepee has debayered it - there shoud be a pattern of squares of RGB pixels):

So believe it or not, converting a RAW dark to a TIFF has applie3d some pretty huge changes.

But DSS works direct with the RAW image - this is what a stack of the frames in DSS gives as a master dark (note the bayer pattern which must come from the white balancing applied by the camera):

This is what I would expect a dark to look like. So I wouldn't worry.

[kalasinman]

Your dark is misleading it looks like you have resampled at 500% and 'smeared' the pixels. You need to zoom in past 100% so you can see individual pixels and do a screen grab.

This is what one of my darks (which looks black with isolated dots as a RAW) looks like when saved by RAW Therapee as a tiff and zoomed in. Without zooming it looks just like yours in RAW Therapee (note that RAW Therepee has debayered it - there shoud be a pattern of squares of RGB pixels):

So believe it or not, converting a RAW dark to a TIFF has applie3d some pretty huge changes.

But DSS works direct with the RAW image - this is what a stack of the frames in DSS gives as a master dark (note the bayer pattern which must come from the white balancing applied by the camera):

This is what I would expect a dark to look like. So I wouldn't worry.

If camera white balancing is applied, it isn't truly a RAW. RAW is defined as pure data without modification. There is a setting in DSS to apply white balance or not, either from camera or custom. Not being the best.

[Stub Mandrel]

If camera white balancing is applied, it isn't truly a RAW. RAW is defined as pure data without modification. There is a setting in DSS to apply white balance or not, either from camera or custom. Not being the best.

I've got an ancient Canon 10D, I can't find a way to turn it off :-/

[KyleStoke]

My opinion for what its worth, i suspect it is stretched, this is what a perfectly fine single dark frame of mine looks like when its been stretched

Have you tried stacking multiple darks in DSS and seeing what the master comes out like

[wxsatuser]

I've got an ancient Canon 10D, I can't find a way to turn it off :-/

You cannot turn it off in camera.

Normally the white balance is applied directly to any jpeg but it's not applied to the RAW.

It's only written to the metadata in RAW.

[Stub Mandrel]

It's only written to the metadata in RAW.

Much headscratching over two poorly written manuals and I think you are right - so why do the darks DSS produces show a slight Bayer pattern (see above)? Light leakage?

If no processing was applied I would expect pixels to read the same whatever colour they are in a dark.

[wxsatuser]

I don't see that when I have used darks but then I don't use darks as a rule.

May be someone else can explain what is going on.

[Hackleby]

A big thank you to all who took time to respond to my plight. Your feed back stopped me from giving in and underlined that I indeed had, yes had a problem.

Having never worked with dark frames before undertaking the trials of Astrophotography my knowledge of them was zilch.

Comments made above with regards the dark frames being stretched puzzled me as even just viewing the RAW files highlighted the same issue and there in lay the problem.

Today I was asked how I was viewing the darks and the resulting conversation opened my eyes. Having received both my cameras without software I had been opening my RAW files and viewing them with the only software I had to do the job, Corel Paint X7 and Picasa 3. I have since installed Digital Photo Professional and taken another look at my RAW Darks frames, they are pretty much totally black and have no resemblance to the images displayed when viewed with the afore mentioned software. 

The tutorial frames how ever do still look similar but just not as pronounced.

Moving forward I believe that I now have some setup issues to resolve in DSS to get the best out of it and try to retain some of that lost colour and detail.

Once again I thank you all.