of flats and darks...

joshman (Josh) · #1 03-09-2007, 07:00 PM

hey, just a quick question about image processing. i plan on doing quite a bit of wide field astrophotography from a static tripod mount, and because of this i need to take lots of short exposures and stack them, and then process them. my questions, therefore, are:

1.) what is a dark, and how to i create one with my dslr?
2.) what is a flat and how do i create one?
3.) are there any tips and tricks for stacking the images?

Alchemy (Clive) · #2 03-09-2007, 07:14 PM

1. A dark frame is one taken at the same exposure and temperature as the shots you are taking, this is subsquently subtracted from your image. but for short exposures ie less than 30 secs i probably wouldnt bother.

2. flat is a shot taken with a lightbox or similar to show dustmotes vingetting etc. varios programs will do the manipulating for you.
a simple and free way of stacking images is to convert your files to a bitmap or tiff and stack in registax, (if its being tracked with no rotation)

a couple of usefull books to explain the nitty gritty is THE new ccd astonomy by Ron Wodowski, and The handbook of astronomical image processing by richard Berry which also has software included that will do all those things discussed

Alchemy (Clive) · #3 03-09-2007, 07:16 PM

Oops forgot to say dark frame has lens cap or scope cap on so it measures the electronic/heat noise.

jase (Jason) · #4 03-09-2007, 07:23 PM

http://www.iceinspace.com.au/index.p...63,211,0,0,1,0

Enjoy.

Dennis · #5 03-09-2007, 07:27 PM

Hi Josh

Alchemy has covered some good stuff above.

Simplistically put, DSLR’s are optimised for day time “happy snap” photos, where exposures are usually sub 1 second duration.

Astrophotography often requires exposures of 30 seconds to 5 minutes duration to record sufficient details in the (dim) subject. At these long exposures, the ccd or cmos sensor begins to accumulate thermal noise, which becomes visible as hot pixels in the image.

So, your image is now made up of the following components:

The subject matter e.g. a galaxy. (=signal)
Hot pixels or thermal noise generated by the long exposure. (=noise)
Optical defects in the system such as vignetting, dust spots, light gradients, etc. (=noise)

Subtracting a dark frame removes the thermal noise.
Dividing a Flat Field Frame removes the system's optical defects mentioned above.
And voila, you are left with the actual “signal” (=galaxy) and the “noise” (darks & flats) has been minimised to an acceptable level.

Cheers

Dennis

joshman (Josh) · #6 03-09-2007, 07:54 PM

ok...so where, if at all, does the onboard long exposure noise reduction that canon dslr's come with come into play, my understanding is that when used, it does the exposure of say, 20seconds, then does a dark straight after and does the subtracting right then and there, it's seemed to work pretty well so far, does anybody have any experience with these, as in which is better to use?

[1ponders] (Paul) · #7 03-09-2007, 08:40 PM

The onboard darks can do a better job than the standard dark subtraction proceedure (not always though) as that dark is optimised for the temperature of operation(though there is some non-random noise that it doesn't take into consideration). However it is time consuming. If your capture time frame is limited to an hour due to a front bringing cloud with it you want to spend your time taking your valuable light (signal) shots and then take your darks when your finished. If time is not a worry then you can take your darks using incamera dark subtraction. This is one of the benefits of active cooling in cameras. You can take your lights at a given temperature and then take your darks whenever you want to by simply cooling the camera to the desired temperature. ( well that's the theory anyway

)

joshman (Josh) · #8 03-09-2007, 08:49 PM

well i did some test images to stack over the weekend, but the exposure time was insanely too long (25 seconds) and as such there is some star trailing and too much other crap, not to mention the minute or so that it took to take each photo with the in camera noise reduction, so next time i'm headed to a place with little light pollution (this weekend actually) i plan on takign much shorter exposures, and a lot of them. i reckon about 20 or so photos at 10 seconds should suffice...yeah?

in camera darks might be feasible for that, i would suppose.

[1ponders] (Paul) · #9 03-09-2007, 08:59 PM

What sort of focal length and f ratio are you operating at? The sorter and faster the better. 50mm at f2.5 would be great. You could shoot for 15 sec for each shot and take 40 or 50 of them then easily. The offset from star movement won't be too bad for that period of time. About 25 min for 50. In that time the sky will have moved around 7 deg. Using a 50mm lens you will have a field of view of 25 deg x 16 deg with a 300D. Plenty of space to crop out the edges.

[1ponders] (Paul) · #10 03-09-2007, 09:02 PM

71 deg x 47 deg for the 18mm f3.5. this will allow you nealy 40 sec of imaging time per shot.

joshman (Josh) · #11 03-09-2007, 09:22 PM

hmmm....i must be on a faster part of the earth, or else my focusing is better than i thought, but i can notice star trailing on my exposures over about 20-25 seconds at 18mm f3.5.

it'd be nice to get the 50mm f1.8 prime lens. that should allow some pretty nice shots.

i'm sure i'll work out a system that'll work well for me.

RB (Andrew) · #12 03-09-2007, 09:47 PM

Josh are you using a cable release and mirror lockup?
If not then camera shake may contribute to your star trails.

Also my advice is to stick with "in cam n/r" with your short exposures until you're ready to start tracking, then think about Darks/Flats.
If you take a 25 sec shot then it will need another 25 sec for the in cam N/R.

There's a limit to how short an exposure you can go to avoid trails no matter how many lights you take, for example 5 sec @ ISO 1600.
The sensor just doesn't have enough time to record enough signal to overcome sensor noise.

So there's many factors that come into play, but without tracking you're limited and at the mercy of these factors.

[1ponders] (Paul) · #13 03-09-2007, 09:49 PM

Or even try your timer so you are well away from the camera when it goes off.

Dennis · #14 03-09-2007, 09:52 PM

Quote:

Originally Posted by joshman

hmmm....i must be on a faster part of the earth, or else my focusing is better than i thought, but i can notice star trailing on my exposures over about 20-25 seconds at 18mm f3.5.

it'd be nice to get the 50mm f1.8 prime lens. that should allow some pretty nice shots.

i'm sure i'll work out a system that'll work well for me.

There will be more star trailing evident if you take long exposure photos in the equatorial regions than the polar regions, due to the larger diameter circles of celestial latitude towards the celestial equator.

Cheers

Dennis

joshman (Josh) · #15 03-09-2007, 09:57 PM

Quote:

Originally Posted by Rocket Boy

Josh are you using a cable release and mirror lockup?
If not then camera shake may contribute to your star trails.

Also my advice is to stick with "in cam n/r" with your short exposures until you're ready to start tracking, then think about Darks/Flats.
If you take a 25 sec shot then it will need another 25 sec for the in cam N/R.

There's a limit to how short an exposure you can go to avoid trails no matter how many lights you take, for example 5 sec @ ISO 1600.
The sensor just doesn't have enough time to record enough signal to overcome sensor noise.

So there's many factors that come into play, but without tracking you're limited and at the mercy of these factors.

i've a cable release, and use mirror lock up, and even leave a bit of time between mirror lock up and opening the shutter, and i'm pretty careful to not even breathe in teh direction of the camera during exposure (slight exageration). so it's most definitely star trails that i'm recording

Quote:

Originally Posted by [1ponders]

Or even try your timer so you are well away from
the camera when it goes off.

i used to do that with the powershot S3IS

i think it'll come down to actually getting out there and experimenting.

[1ponders] (Paul) · #16 03-09-2007, 10:20 PM

Hang on I just thought of something. If you are using an 18-55 lens with the 400D then the 18mm is infact giving you around 28.9 mm lens. This will only allow 25 sec of exposure. Sounds like you need to take a journey up the coast to try some lenses and tracking

RB (Andrew) · #17 03-09-2007, 10:36 PM

Unless you can go ultra wide, say 10mm-12mm, you'll need tracking to avoid trailing and get a half decent shot. Even then tracking will eventually be required for great ultra wide shots.

joshman (Josh) · #18 03-09-2007, 10:54 PM

Quote:

Originally Posted by [1ponders]

Hang on I just thought of something. If you are using an 18-55 lens with the 400D then the 18mm is infact giving you around 28.9 mm lens. This will only allow 25 sec of exposure. Sounds like you need to take a journey up the coast to try some lenses and tracking

is that an informal invitation? i hear that there's a bloke up north with a pretty big lens... that's begging for a 400D

[1ponders] (Paul) · #19 04-09-2007, 07:32 AM

You're welcome to come up anytime, just not this Saturday as I'll be out at Barambah for Saturday night. Drop me a pm if you are interested and I'll send you my address.

[1ponders] (Paul) · #20 04-09-2007, 07:32 AM

Or we could go up and play at the Mapleton Observatory.