understanding dark frames

[Doc]

I'm finding it hard to get my head around dark frame subtraction.

To get dark frames I just block light coming into the telescope is this correct? And the total exposure of dark frames must equal the total exposing time for the given object?

I understand the theory behind noise etc but isn't dark frame subtraction essentially like using your in camera noise reduction or a third party software such as noise ninja

Also when are the dark frames subtracted?

Mick

[jonifive]

Good question.......any kind of answer here would be a great help as I must admit to knowing nothing about dark frames. Even google was a bit vague!

J

[Doc]

Maybe some really knowledgeable person could write up an idiots guide to dark frames

[moondog]

Doc

Yes, for dark frames you block the light to the scope. If you take say 10 light frame subs of 45 secs @ 400 ISO you take the same number and exposure of dark frames. (At the same temperature - this is important)

In my experience dark frame extraction by software is superior to in camera. Same time to take subs, but you can do the dark frame subtraction after you've finished taking all the light frames. Cuts down on possible cloud cover spoiling the light frame sequence !

Noise Ninja (which I use) is excellent, but you get better results using N N after dark frame subtraction. Most astro software will automatically subtract the dark frames from the light frames for you.

One other useful sub to take is a bias frame as this deals with any inherent noise in the CCD or CMOS chip. This is a sub taken with the lens cap on, but at the highest camera speed. One sub is sufficient.

MD

[jonifive]

You lot are all mad with your yin yang astrophotography!

I've got a perfectly good camera and you want me to take photos with the lens cap on!!!

[Astroman]

Finally, an image process I know well. First though, the only image processing I do is bias subtraction, dark frame subtraction and flat fielding for science. No other color rendering or anything. (I don't take pretty pictures, I just take deep images of random, not-too-interesting star fields. That's why I don't post any.)

You take a bias frame and subtract it from your master dark frame to establish a baseline or benchmark for the digital numbers which make up an image. With this benchmark established, it's then possible to compare directly images I take with my less-than-professional grade camera to one taken with the Gemini North 8 meter telescope on Mauna Kea. Flat fielding is the opposite of dark framing and removes any light objects from my image, eg dust donuts or unevenness in illumination of the field.

This is where I stop processing and submit data to AAVSO and so on. This is where masters like Celescope take over and make stunning, balanced images.

HTH.

[Kaptain Klevtsov]

So to add into this the theory behind dark frames...

Every camera will, even with the lens cap on, form an image composed of pixels that become lighter than the black background because of faults in the imaging sensor. The level of this noise is related to the temperature at which the sensor is doing the imaging because its a function of non-visible infra red light from the internal camera workings and dodgy bits in the imaging sensor itself.

During a long exposure image taking session, each frame will be affected by the same noise that would affect the image with the lens cap on. To combat this we do two exposures. One is the target image and the other is with the lens cap on.

By subtracting the lens capped image from the sky image it is a simple matter of removing the false bright pixels from the final image and you then get a better view of whatever you want to mage.

Kaptain Klevtsov

[Ant]

So I have a question on this then.

Say I do 50 x 1 minute exposures. In the past I would then do a dark frame of 1 minute - but just one of them.

Then do the subtraction.

Do I really have to do 50 dark frames?

If I do have to do 50 dark frames, what do I do with all of them - stakc them in the same was at the light frames? then subtract?

Ant

[Doc]

Good question Ant which I would ike to add another to.....

If like Ant said he had to do 50 dark frames why can't he just do one master and copy an extra 49 in photoshop.

Is the same as covering your telescope lens 50 times.

One more question.

Are the 50 lights and 50 darks added together into one photo then subtracted?

[daz]

The reason for taking the same number of darks is that the readout values will vary on each exposure according to length. With a DSLR you need to do darks each time. With a cooled CCD, you can build a library of darks at 'x' exposure length at 'y' temperature.

The darks all have the bias subtracted, then averaged together to produce a master dark which are then subtracted from each light.

Flats fields do not need as many exposures, but if you are using filters, you will need a flat for each filter. If you move the camera, you will need flats for the new position, etc. The flats are then averaged together - by the separate colours if necessary - and subtracted from the lights - of course, if you have RGB images, you subtract the R flat from the R light, etc

Its all good stuff

[Robd]

Just to throw my two pennyworth in - I suspect it raise more questions than answering any

I only take a single dark at each exposure that I have used during an imaging session. This is mainly due to the fact that the stacking software that I use ( ImageSXM on a mac ) will only process 1 dark per set of images. If temerature is changing much then I will take more. This would apply mainly to those dark winter nights when you get out at 20:00 and stay till 02:00 and beyond. The temp. change can then be quite significant.

I do not know whether in camera subtraction is worse or not, but it will obviously double the length of each shot. Which is a major overhead for me since I only need 1 dark.

Assuming that the principle of the same number of darks as lights is correct it is still better to do darks rather than in camera subtraction since you can take the darks at the end of a session, while a cloud passes etc. You also have the advantage of the unpprocessed images to play with.

Another thing to bear in mind when taking a series of images with a DSLR is that the warmer the ccd gets then the more noise. This is affected by the ambient temperature and the duration of the shutter open time, hence the longer the exposure then the more noise. But remember that to some degree this is cumulative. If you take say 10 2min exposures with no break between them then the electronics have no chance to cool between exposures. There will then be a gradual increase in the background noise, especially that generated by the amp.

This effect will vary from camera to camera. My K10D is not to bad, my *ist D was c**p.

Ant - re your question about stacking the darks then I would assume taht yes stack then and then use the result as the dark. Some software ( DSS ) may well do it for you - 10 darks, 10 light, 1 flat - get on with it

Doc - No the dark ( stacked ? ) wil be applied to each light as it is loaded otherwise you are potentially applying the noise reduction to the wrong bit of the image - stacking will have aligned the light images and therefore shifted them all a little. This process is also removing hot pixels. I do not know how this is accomplished in anything other than ImageSXM where the hot pixels are identified from the dark and stored in an array. As each image is processed ( including the dark ) then the hot pixel is replaced with the average of the same colour around it.

Again this has to be applied to each image.

Re take 1 and copy 49 - the noise generated will be slightly different for each exposure. Adding the 50 copies together and averaging will just take you back to the origional ( 1 x 10 / 10 = 1 )

Enough rambling - time to lie down in a dark room....

Rob

[Doc]

Wow this has frazzled my brain.

It's most proberly easier then it sounds, love reading about it but as for doing it we will see.

[barkis]

Where are the temperatures measured from,? has the camera got an in built temp probe which puts a reading on to the screen?

It is obvious I suppose that the temp of the sensor is the only one that matters, therefore ambient temperature is not affecting anything.

I own a dslr, but all this info is now important to me, as the camera will be playing a big role in my astro imaging.

Some of these remarks may seem like statements, but I am really asking questions.

Ron.

[moondog]

Ron

The ambient temperature of the air will affect the CCD or CMOS chip. Usually in Winter the noise produced by the camera is far less than in the higher temperatures of Summer (!). I can often get away with no darks in Winter or when it's cold, but need them in Summer when it's warmer. I don't think any camera has a built in temp. probe.

MD

[barkis]

Ron

The ambient temperature of the air will affect the CCD or CMOS chip. Usually in Winter the noise produced by the camera is far less than in the higher temperatures of Summer (!). I can often get away with no darks in Winter or when it's cold, but need them in Summer when it's warmer. I don't think any camera has a built in temp. probe.

MD

Thanks MD.

[Vega]

Now realising that my methods of taking 1x30 sec dark frame for a whole load of 30 sec light frames is flawed.....

I'm gonna go back to turning on my 400D's auto noise reduction and be done with it. Putting a cover on for every frame I take is just too much hassle for me. I spent out on ImagesPlus software so that I can pop in and have a cuppa while my camera is snapping at DSO's [sorry I.. know... no commitment im gonna get blasted for this!]

Vega

[jcm]

Now realising that my methods of taking 1x30 sec dark frame for a whole load of 30 sec light frames is flawed.....

I'm gonna go back to turning on my 400D's auto noise reduction and be done with it. Putting a cover on for every frame I take is just too much hassle for me. I spent out on ImagesPlus software so that I can pop in and have a cuppa while my camera is snapping at DSO's [sorry I.. know... no commitment im gonna get blasted for this!]

Vega

I think one dark for for a whole load of images is not enough.

BUT I do not take one dark for every image. If I have taken say 30 images I might take say 10 dark frames. Most processing software will create an average single dark frame ( masterdark) and use this when subtracting from the frames. This seems a good compromise between time imaging and time taken to improve the quality of raw data.

John

[barkis]

Surely a dedicated CCD camera, with Peltier cooling down to minus 25, or whatever temp. they get down to, cancels the need for so many dark frames. I haven't a clue, as I am just getting into this new exciting aspect of astrophotography. I really would be a bit miffed if I am taking frame upon frame of dark images, just to eliminate a few hot pixels from your final image. It's a bit laborious isn't it?

And, as my recently acquired camera is a MX916 mono, I need to do more if I want to create a colour image of say M42, as each exposure is repeated using the different filters. The more sensitive chip appealed to me, but taking multitudes of dark frames doesn't.

Oh well, let's get on with it.

Ron.

[SteveL]

Putting a cover on for every frame I take is just too much hassle for me.

err... I just take my darks as a completely seperate sequence after my lights. No need to take a dark after each light.

[Kaptain Klevtsov]

I'm with Ron on this Re. CCD astrocams. I don't do darks with the Atik, though they were absolutely necessary with the SC3 webcam. The DSLR also needs darks but I tend to just hoover up the hot pixels in Photoshop at the moment as I rarely get the flippin' thing to work. Some time to practise all this would be good so somebody make the clouds go away. Please?

Kaptain Klevtsov