Noise reduction in DSLR camera

[petergc]

Morning all. I took advantage or a clear spell last night, popped out to my light polluted garden for a crack at Orion with my 60d DSLR. Went okish and I will share image later.

Now I have a noise reduction option in camera, is it wise to have that up high should it be turned off? Pros cons if any.

Thanks

[wxsatuser]

If your just taking one exposure then on would be ok.

If your doing many subs better off and do darks etc.

[petergc]

Thank you wxsatuser. I will have to go at it again.

[baggywrinkle]

I am trying to find an article I came across saying that the noise reduction setting when used in astro-p work can cause the image to be worse. Better to use lights & darks. Seems odd but it did stick in y mind.

[Sam]

I think it is better to disable the noise control and do you own through processing. That way you've got complete control over what's going on.

[ronin]

Really a difficult choice. Allowing the camera to take it's own "dark" and do the processing internally is straight forward and makes life easy.

However as this means the chip is powered on for say 60 seconds to get the image then a further 60 seconds for the dark the thing has power to it for a 120 second period and so more heat produced, which is best if avoided.

If just getting images the often recommended that the image is taken then the camera/chip is left for the period of time to cool dawn. This does not happen with the camera doing it's own noise reduction process.

I would say that as a first or initial go at AP then let the camera do it. This would then double the time between exposures. For the simple reason you may not be ready to take your own dark, then add them to the processing software and incorporate them in easily. Everything and everyone will I guess say it is easy, and it often is if you haver done it for the last 2 years. However as a first attempt I find that the reality is often different.

Digital cameras are easy: You try something and see what the result is, if good then keep it, if poor then delete it. Then when someone else asks the same question in 6 months you can report your experience and findings.

[wxsatuser]

I have done single exposures both ways and to my eye could'nt see the difference.

When I do lots of subs its definately turn off in camera noise reduction.

[Zakalwe]

If you enable the noise reduction in camera, then after each exposure the camera will take a dark frame of the same duration. So if you are doing 5 minute subs you will only get 5 minutes of usable data every 10 minutes.

In other words, in camera dark frame subtraction will loose you 50% of imaging time. That might be berable if you lived in Arizona where clear nights are ten-a-penny, but not in our rotten climate.

Its much, much  more efficient to gather your data and then do 20-25 dark frames at the end,. Then let your stacking process (DSS for example) do the dark frame subtraction.

[Alien 13]

I agree with the loss of valuable imaging time using the inbuilt system but i have wondered if the auto darks are more accurate i.e. they are taken with exactly the same conditions frame by frame.

Alan

[Zakalwe]

I agree with the loss of valuable imaging time using the inbuilt system but i have wondered if the auto darks are more accurate i.e. they are taken with exactly the same conditions frame by frame.

Alan

I don't know Alan. I personally have never tested it. I have seen posts saying that the temperatures don't have to be exactly the same...a few degrees either way don't make that much difference. And if there is a difference (which would probably be marginal) I am fairly confident that the difference wouldn't come close to the improvement that a whole load more sub-exposures would make.

If noise control was that crucial then a cooled CCD is leagues ahead of a DSLR. I personally know which one I would go for...

[wxsatuser]

What I understand is that darks can be taken inadvance, at about 10degree intervals, over the range of temperatures you are liable to image in.

The processing software will need to be able to scale these darks to suit the exposures.

Also it takes approx an hour for a dslr to stabilise it's internal temperature, so take a lot of darks at the iso and length you use and throw away the first hours worth.

Of course you don't have to go to these lengths if your not to fussy on the results.

[thunderchild]

I do my dark's in the fridge, i leave my 1100d in there for a few hours then start a plan in apt and then let it run. I have made up a large library doing it this way and use the dark's that are +/- 3 degrees either side of my image temp,  and it gives me more time to capture more data in the field which is always a good thing.

[SnakeyJ]

The inbuilt dark feature is taking and subtracting a single dark frame after each exposure, which as already stated will waste half of your imaging time.   Though its far better to take a series of darks to average out the effect, creating a master dark frame - typically I take 15-20 darks for each iso/exp/temp for my library.   The fridge method is good, though mine are taken on rainy/cloudy nights in the observatory which is suitably cold and dark.    I use APT, which lets me take a series of different exposure/iso runs unattended overnight.

[IanL]

The biggest problem with DSLR darks is that the camera (this applies Canons, I haven't tested other makes) processes the light frame prior to writing out the RAW file.  It reduces the appearance of dark current but not dark current noise, which makes for better daylight pictures. Doing so hinders efforts to apply CCD camera techniques like dark subtraction to DSLR images.  (More here from Craig Stark http://www.cloudynights.com/item.php?item_id=2786).

Some have suggested to me that the result is that it is pointless to use Darks on DSLR images, but my experiments lead me to think otherwise.  It is not as easy as it could be to make them work though:

- Make a master bias first (30+ dark frames, which are shot at the camera's shortest exposure and then stacked).  If you stretch the master bias you will see that there is usually a pattern of vertical stripes 1 pixel wide.

- Make a master dark by stacking 30+ frames that are within about 3 degrees C of the main exposure.  So if my exposure has an EXIF temperature of 11 C, I'd use a master dark stacked from about 30 dark frames with EXIF temperatures of 10, 11 and 12 C.

- Subtract the bias from the master dark.

- Subtract the bias and the master dark from the light frame 

(The alternative is to forget the bias altogether, as the master dark contains the bias and you could subtract both from the light frame in one operation but that is not what I do.)

- If you are doing flats, you can usually just subtract the bias from the flats and stack them since the exposure is so short the dark current won't have built up enough to matter.  You divide the light frame by the master flat.

- Then stack your lights.

Making a master dark is pretty essential really.  A single dark frame taken and processed on camera has a lower signal to noise ratio than a master dark made from multiple darks stacked together. When we talk about signal here, we are talking about dark current, and noise is the dark current noise and readout noise combined.

Bear in mind that using a calibration frame of any sort always increases the noise in the image, so you need it to have a good signal to noise ratio to make it worthwhile (where signal is whatever we are trying to correct, fixed pattern - bias, dark current - darks or differences in response - flats).

The other problem with an on-camera dark is that you can't really be certain the single dark is even close to the temperature of the light frame, especially at the start of the imaging session where the camera is not at thermal equilibrium (if indeed it ever is!)

[Uranium235]

Fascinating post from Craig Stark, well worth reading. Especially in regard to ISO settings.

[petergc]

Thanks very much for all replies. A lot for a beginner to take on and with that said I will continue over and try to make some sense of the above link.

[SnakeyJ]

Sorry Peter, there's lots of great info in Ian's post and the Craig Stark articles - though it is a very steep curve, especially when there's so much else to learn at the same time.

Actually these sensors are pretty good - and the main thing is to capture some lights to stack. You can always come back and re-process with bias and darks later on. We get the odd night or two when the weather keeps us in - so always time to experiment.

Always worth a browse through Steve Richard's book MEPC - a firm recommendation from all of us here.

typed on my mobile with Tapatalk

[IanL]

Sorry Peter, there's lots of great info in Ian's post and the Craig Stark articles - though it is a very steep curve, especially when there's so much else to learn at the same time.

Actually these sensors are pretty good - and the main thing is to capture some lights to stack. You can always come back and re-process with bias and darks later on. We get the odd night or two when the weather keeps us in - so always time to experiment.

Always worth a browse through Steve Richard's book MEPC - a firm recommendation from all of us here.

typed on my mobile with Tapatalk

Absolutely right.  After learning to stack multiple light frames (which will make a massive difference to your images), flats will make the next biggest improvement to your images so tackle that second.  You can always make darks and bias later and re-process existing images with them, but you can't really take flat frames after the event as the camera orientation, focus and the bits of dust and fibres on the sensor all need to be the same between light frames and flats.  I have some nice images covered in dark dust shadows that I'll never be able to rescue due to problems with flats.

[Physicist13]

just bear in mind that if you take darks at a later time, try to get have the camera at approx the same temp as when the lights were taken. The dark current scales quite dramatically with temp. That being said some processing tools can scale darks at slightly different temps to match the temp of the lights if they differ (PI for example)

[wxsatuser]

just bear in mind that if you take darks at a later time, try to get have the camera at approx the same temp as when the lights were taken. The dark current scales quite dramatically with temp. That being said some processing tools can scale darks at slightly different temps to match the temp of the lights if they differ (PI for example)

ImagesPlus as well.

[Knight of Clear Skies]

Thanks for the informative posts, this is something I've been wondering about as well. I'm only using a fixed tripod at the moment, so I think I'll try turning it on for single exposures, but will turn it off when taking a set for stacking and take my own darks.