Canon 6d Does not need darks - discuss

[powerlord]

So, searching the net this seems to be controvertial at the time. Has opinion solidified one way or the other ?

I'm referring to Clarkvisions in depth test and review of the original 6d (which I have) btw, not my own opinion - I'm just wondering if it's true not, as clearly it would save me a load of hassle and time...

https://clarkvision.com/reviews/evaluation-canon-6d/index.html

Specifically the summary paragraph:

" The constant dark level with long exposure time indicates the camera has on-sensor dark current suppression. This, however, does not suppress noise from dark current. But it results in a uniformly dark level that needs no post processing correction. No long exposure dark frames are needed when making long exposures if recording raw."

It does sort of tie in with what I was finding when I forgot and didn't have time to take darks - I even reprocessed an old shot or two without them to see if there was any difference, and then wondered what I was doing wrong when I couldn't see any.

Note: do read the article - he's not talking about not using darks if you turn on long exposure noise reduction and/or high ISO speed NR. He's saying with those off, there's no point in taking darks - since whatever thermal dark current noise is there at the temperature - is random so you ain't getting rid of it - and the camera does the rest.

Also, fascinating the difference in thermal noise with temperature. Really brings it out why folk use cooled CCDs! Look at those 1C pics !

[Roy Foreman]

Interesting.  I use a Nikon D810a, a remarkable DSLR designed specifically for astro imaging. I do not generally need to take dark frames with it, the noise levels are so low.

A couple of years ago I moved to cooled CCD cameras thinking that they would produce lower noise levels with shorter exposure times. Not so. I have yet to produce an image with these cameras that matches what I can achieve with the Nikon. More specifically, 60 min of exposure on M51 with a ZWO 294 MC produced more noise and less detail than 15 min with the Nikon. With the Nikon it takes generally 10 - 20 frames to reduce noise levels to acceptable levels. with the ZWO it takes at least 60 !

I guess companies like Canon and Nikon have huge resources and technology to achieve this.

 

[michael8554]

If it's cloudy tonight you'll have plenty of opinions to read on the subject of DSLR Darks with different temperatures !

I load Bias subs as Darks in DSS with my 6D.

Michael

[powerlord]

huh. so you just shoot lights, and point DSS at your bias master as a dark ? and that works well for you I assume. Might give that a try then.

stu

[michael8554]

Forgot to say - you must Dither too !

Michael

[rnobleeddy]

I don't think many people recommend darks for DSLRS anymore - rather, just dither as much as you can each frame and use a bias.

I haven't compared side by side, but the evidence/consensus is pretty solid - it's impossible to match the temperature. Once you factor in the hassle/time used to take darks, I can't see any possible advantage.

Good to know the 6D is low noise though - considering picking one up!

 

[david_taurus83]

I've made a couple of DSLR cooler boxes and created quite an extensive dark library for my 6D by 'matching' EXIF temps by controlling the ambient temperature around the camera. I can't see any real difference between using darks versus not using them. The 6D is very low noise already. It produces very clean stacks though I have only ever used mine from my light polluted back garden so when I remove the LP it does look noisy in areas of low signal. I would concur that not using darks and to dither is more beneficial. You must use a master bias though. The 6D bias is a whopping 2047 ADU out of a possible 16k ADU. It gives the impression that your lights are more washed out than they actually are so I try to expose until the histogram (in software, not on camera) is around the halfway mark.

[Ken82]

Ive always dithered and used darks with my canon 6d. Its an excellent camera even to the point I much preferred it to my asi 071.

Ken 

[powerlord]

3 hours ago, david_taurus83 said:

I've made a couple of DSLR cooler boxes and created quite an extensive dark library for my 6D by 'matching' EXIF temps by controlling the ambient temperature around the camera. I can't see any real difference between using darks versus not using them. The 6D is very low noise already. It produces very clean stacks though I have only ever used mine from my light polluted back garden so when I remove the LP it does look noisy in areas of low signal. I would concur that not using darks and to dither is more beneficial. You must use a master bias though. The 6D bias is a whopping 2047 ADU out of a possible 16k ADU. It gives the impression that your lights are more washed out than they actually are so I try to expose until the histogram (in software, not on camera) is around the halfway mark.

thanks - good info there.

I assume you also use flats though ? if nothing else, to minimise dust/lens marks, etc?

[david_taurus83]

3 hours ago, powerlord said:

thanks - good info there.

I assume you also use flats though ? if nothing else, to minimise dust/lens marks, etc?

Yes flats are always a necessity. I just use a master bias as a flat dark.

[Chefgage]

On 29/04/2021 at 23:58, rnobleeddy said:

I don't think many people recommend darks for DSLRS anymore - rather, just dither as much as you can each frame and use a bias.

I haven't compared side by side, but the evidence/consensus is pretty solid - it's impossible to match the temperature. Once you factor in the hassle/time used to take darks, I can't see any possible advantage.

Good to know the 6D is low noise though - considering picking one up!

 

I did a side by side comparison using my canon 200d. Processed both stacks (one did not have the darks included in the stacking). Processed both side by side using the exact same process. The result is I could not tell the difference between the two. So for me, my conclusion was I don't bother with darks any more, means I can get to bed quicker now

[alacant]

On 30/04/2021 at 11:32, david_taurus83 said:

not using darks and to dither is more beneficial.

+1. AFAICT that applies for any modern dslr. The eos450 era and before do however seem to benefit from dark frames, if only to help remove banding. But it's 2021 and well...

Oh, and don't forget to stack using a clipping algorithm.

On 30/04/2021 at 18:31, david_taurus83 said:

use a master bias as a flat dark

Perhaps best to use a master bias as a master bias. We've not seen any benefit using dark frames of any type, be they dark for flat or dark for light.

Just our €0.02  but HTH anyway. 

Cheers

Edited May 4, 2021 by alacant

[david_taurus83]

1 hour ago, alacant said:

 

Perhaps best to use a master bias as a master bias. We've not seen any benefit using dark frames of any type, be they dark for flat or dark for light.

Just our €0.02  but HTH anyway. 

Cheers

That's what I meant. I use Pixinsight to manually calibrate. I remove the bias from the flats prior to creating a master flat.

[alacant]

On 29/04/2021 at 09:54, powerlord said:

referring to Clarkvisions

I believe some (most?) of his reviews are nonsense.

HTH

Edited May 4, 2021 by alacant

[vlaiv]

Not using darks means improper calibration.

Many people think that dark subtraction has something to do with noise - no it does not. It has to do with signal.

You need to remove dark current signal if you want your flats to work properly. Only light entering thru telescope is subject to vignetting and dust shadows. If you don't remove dark signal - you'll end up trying to correct something that was not subject to vignetting and dust shadows in the first place - and you'll have opposite effect - you'll "over correct".

Point of calibration is to exclude all the signal that should not be in the image and leave only signal from light that reached sensor thru aperture (if there is any other kind of light present - well, then you are in trouble as you have light leak).

One can properly calibrate with darks that are not temperature matched if temperature was recorded, dark doubling temperature is measured and known and bias is stable on particular sensor. Even if temperature of darks is unknown and doubling temperature is not measured - there is algorithm called dark optimization that tries to guess scaling factor for darks. Only requirement is stable bias.

[powerlord]

20 minutes ago, alacant said:

I believe some (most?) of his reviews are nonsense. 

HTH

ah. thanks - good to know.

still, concensus (well nearly) seems to be don't bother  - just use bias + flat. So that'll do for me frankly. I ain't a pixel peeper - as Lazy Geek on youtube would say.

[alacant]

4 minutes ago, powerlord said:

just use bias + flat.

Exactly.

In the real world, who cares about theory? Especially when it just works.

Summary for modern dslrs:

-bias flat and light frames

- dither between the latter

- stack with a clipping algorithm.

That's it.

[vlaiv]

2 minutes ago, alacant said:

In the real world, who cares about theory? Especially when it just works.

All those people that designed and built device you use to leave this comment and all those people that designed and made equipment that we are discussing for example.

I bet they all care about theory quite a bit.

[alacant]

3 minutes ago, vlaiv said:

people that designed and built device

So a tiny minority (if any) of those who use this forum;)

Many of us here want to know what works. Tips if you will.

Cheers

[Seelive]

4 minutes ago, alacant said:

... Many of us here want to know what works ...

I suppose their will always be both acceptable and optium methods for anything.

[vlaiv]

19 minutes ago, alacant said:

So a tiny minority (if any) of those who use this forum;)

Many of us here want to know what works. Tips if you will.

Cheers

I agree, after all, this is a hobby and as such people are free to enjoy it in which ever way they prefer.

However, we should not dismiss benefits that come from proper understanding of the topic. After all - some of people enjoying the hobby do like to have a deeper understanding of the topic and do care about theory of it.

[jetstream]

17 minutes ago, vlaiv said:

some of people enjoying the hobby do like to have a deeper understanding of the topic and do care about theory of it.

Like me, as a wanna be imager (for now). I have fun learning about this subject- which will hopefully allow an understanding of signal collection and data processing for use in practise.

It sure beats wandering around in the dark wondering why things work-or dont IMHO.

Please continue vlaiv

[vlaiv]

5 minutes ago, jetstream said:

Like me, as a wanna be imager (for now). I have fun learning about this subject- which will hopefully allow an understanding of signal collection and data processing for use in practise.

It sure beats wandering around in the dark wondering why things work-or dont IMHO.

Please continue vlaiv

Indeed.

I think it is better to understand the process and then choose not to apply darks than it is to just follow advice without actually taking a moment to understand the principle.

If we just follow a recipe - there will be a moment when things don't work as we expect and in those circumstances having understanding of how things work might help us understand why it's not working in that particular case.

[KP82]

11 minutes ago, vlaiv said:

Indeed.

I think it is better to understand the process and then choose not to apply darks than it is to just follow advice without actually taking a moment to understand the principle.

If we just follow a recipe - there will be a moment when things don't work as we expect and in those circumstances having understanding of how things work might help us understand why it's not working in that particular case.

I'm with you here. I actually enjoy reading many of the theory posts you've made here (seriously). They've helped me in learning about imaging. Please keep them coming.

Edited May 4, 2021 by KP82

[alacant]

1 hour ago, vlaiv said:

 

If we're interested in theory, then please let's check it's correct.

Unfortunately, sometimes it isn't. 

The clarkvision article linked herein, is a prime example.

Thanks