Don't bother with dark frames when using a DSLR

His arguments are weak. Of course you shouldn't waste precious time under dark clear skies to take dark frames. They can be easily taken after imaging session while temperature is still within few degrees from light frames. Or different/cloudy night with the same ambient temperature (or just in the fridge :happy-2. Dark current usually doubles approx. every 5-6 degrees of Celsius.

So as long as you don't waste your imaging time and can match the temperature between light and dark frames - it's always better to make darks and calibrate lights with them.

I agree with Nadir Astro. I usually start my darks right after imaging, so at least an hour before dawn. I also don't have a permanent setup, so when I'm done imaging I bring the scope inside, but into an entryway I've left open to match outside ambient. In addition, I bag the scope in a heavy duty (i.e. thick and opaque) trash bag to further reduce light and moderate temp change. Not completely sure about temp though, the bag might also be trapping heat from the camera….

More importantly for me is that the darks take care of the infamous Canon banding I get with my 5D/IV, especially at particular ISO's. PI has a Canon de-banding script, but so far at least I'v only gotten it to trade gradients for banding. I can more easily take care of gradients during processing, but I'd rather not have to deal with either.

Cheers,
Scott

I've been using a DSLR for some time and it definitly helps to use darks frame. In my case (Nikon D5600) there is a big red/violet amp glow on the bottom of the images that dark frame really help remove, even if they are a few degrees off.

Agree with all here.  My Canon 6D is already pretty good at handling noise (w/o NR on, of course) but on those warmer nights (and as the camera ages, it seems)  I develop more hot pixels and the need for darks is obvious.  I have always used darks and will continue for noise and hot pixels sake, but I have dumped my bias frames as they have always complicated my stretching for some reason.  And yes, building up dark libraries is huge!!

Well said, Arun.  I wish I had met you 4 years ago!  Bias frames certainly do introduce some crazy stuff in my stacks.  Thanks for that expaination.

Bias frames are valid only for CCD sensors and are meaningless and even detrimental for CMOS sensors which are what are used in DSLRs. Depending on the length of the exposure, CMOS sensors can use different timing circuitry (on chip versus off chip) so a bias frame taken at a very short exposure will not be representative of the bias signal in a multiple minute or multiple second exposure. This is why you can get all kinds of strange things happen when you use bias frames with CMOS sensors. The best thing is to calibrate both flats and lights with darks of exactly the same exposure time as the frames you're trying to calibrate.

Yes and no. It depends. On my ASI294MC I had worse results while NOT removing the master bias frame from the master dark frame although the suggestion from ZWO was not to. OTOH, the Nikon D5100 seems indifferent which ever way you do it. I just keep doing it for consistency sake and also because I suspect there is thermal drift in the dark signal which I cannot completely zone in as the actual sensor temperature isn't really controlled or even known.

Andrea:

John Upton did a very detailed quantitative test on the ASI294MC Pro specifically. It goes into specifics of the differences in timing circuitry for different exposures etc:

https://www.cloudynights.com/topic/636301-asi294mc-calibration-%E2%80%93-testing-notes-thoughts-and-opinions/ 

It was this reading that improved my knowledge of how CMOS sensors work and why taking bias frames is not really recommended. I also use a Canon 5D Mark IV on occasion. I had lots and lots of trouble getting rid of banding in my early days when I was using bias frames. Going to only subtracting darks fixed the issue.

These are John's recommendations. They are specific for the 294MC, but the general concept is useful for all CMOS sensors:

"Putting It All Together   

All of my testing has lead me to the following conclusions. I intend to implement these in my usage of the camera if I ever see a string of clear nighttime skies again.

• There are no color channel differences in the Dark Current from the camera


• There is a color channel gradient component in the Fixed Pattern Noise of a Bias Frame


• A Bias Frame directly from the camera should not be used as is for any calibration work


• A manual adjustment of Bias Frames must be performed if they are used for calibration work


• In general, frames at exposures between 0 and 3 seconds should not be used for calibration work


• The camera should be allowed to “thermally soak” for 5 to 10 minutes after reaching the desired set-point temperature


• Thermal stability of the cooling seems more important than extreme cooling. (It is best not to run the TEC at power levels greater than about 80%.)


• Frames of 1 second exposure duration taken between imaging frames helps with session to session consistency"

Hello, this is a very interesting topic, @Stuart Taylor. Thanks for bringing this one up.

Coincidentally, I have also been dealing with the calibration issue for quite some time. I currently use a Canon EOS 80D (CMOS) and calibration has always puzzled me. In addition, you can find so many, sometimes divergent or even contradictory statements on the net on this subject.

I don't know if you are using PixInsight. With the latest release of PixInsight (1.8.8-8 ) a very detailed and up-to-date description of the calibration process (ImageCalibration in PixInsight) by @Edoardo Luca Radice (Astroedo) appeared for the first time. I would even call it an excellent tutorial on calibration in general. Among other things, he discusses when the use of bias frames makes sense and when they are more likely to cause additional problems (with regard to CCD and CMOS sensors, amplifier glow, etc.), when dark frames should be used (calibrated and uncalibrated), dark flats, dark frame optimization etc.

Maybe @Edoardo Luca Radice (Astroedo)'s "tutorial" is available outside PixInsight (if you don't have access to the latest PixInsight release)?

Another very detailed paper on this topic was written by Craig Stark ("Profiling the Long-Exposure Performance of a Canon DSLR" ). He meticulously examines the unexpected behavior of DSLR cameras. Although he mainly deals with Canon cameras, the general conclusion is that it is worth the effort to investigate the behavior of one's own camera, if calibration is taken into account, since general statements are difficult to make due to the apparently diverging behavior of different camera models. In my opinion, Craig Stark's paper describes excellently how to test the behavior of the sensor yourself and what conclusions you can/should draw from it.

Best regards - Thilo

gosh this is really going in some unexpected directions then! It's beginning to look as if I won't need dark frames OR bias frames with my DSLR!

That could save a bunch of time!

I guess I'll just try stacking with and without them and see what works best

@Thilo - no, I am not using Pixinsight (didn't know about it). Is it for stacking? I use DSS and Photoshop

Just stacking the lights definitely does not produce the same starting point for post-processing in my circumstances; typically shooting 30-75s unguided subs in fairly light polluted (Bortle 7) skies. Both with a regular and full spectrum Canon EOS 700D.

For me, I don’t really see the problem. I just take the scope of the mount for shooting darks and bias frames, which gives me ample time to break up the mount and other equipment.

I do feel it matters how long the subs are. For much longer integration times, the relative contribution of read-out noise to the overall noise is different. 

Roger Clark has a rather extended explanation of the different steps here:
https://clarkvision.com/articles/exposure-f-ratio-aperture-and-light-collection/

This thread is really interesting... Thanks for it. I'm reading here since yesterday and I did a test with data I collected last week to have a practical view on it.

I stacked all possible (partly senseless) combinations. If anyone is interested to see the results (full resolution JPGs, 100% quality): https://astrob.in/rue1z5/0/
I didn't upload TIFFs, since they are >100MB each.

For my modded Canon EOS 70D (H-Alpha sensitivity 80-85%) and this particular session data (180s exp.) it seems that I do not benefit from Bias frames, like explained by @Arun H..

I think everyone has to test it for his own equipment, also with different session data...

Thank you for trying even the senseless combinations, this is a really helpful experiment for someone approaching calibration for the first time. For the reasonable combinations, maybe it will be more suitable to apply a low-order ABE (or DBE) with exactly the same settings and then compare. Most people discover improper calibration after removing this thin layer of light pollution from their images.

And I totally agree, all of us should try this experiment!

Whether there is benefit to dark frame subtraction or not will depend on the characteristics of the sensor.

There are two components to dark current - mean dark current, and the noise due to dark current 

Dark frame subtraction subtracts out the mean dark current, but does not, and cannot, subtract out the noise from dark current. The noise from dark current simply adds in a sum of variances manner to the noise from other sources such as read noise and photon shot noise. To reduce noise from dark current requires lowering the sensor temperature, there is no other method. 

If the dark current distribution in your sensor is spatially uniform, then there may not be a benefit to dark frame subtraction, since the mean dark current is simply a constant bias or pedestal that can be handled by tools such as background extraction and color calibration.

If the dark current distribution is not spatially uniform, then you will benefit from dark frame subtraction. Although you could, in theory, handle such non uniform gradients using things like DBE, dark frame subtraction will greatly simplify things.

On my 2600MC Pro, the dark frames look (visually) very uniform, so I could, perhaps, get away without dark frame subtraction. 

My ASI 1600 MM Pro shows glow on one side of the sensor. Dark frame subtraction cleans it up nicely.