Flats, bias and dark frames

[mathewb]

I've been doing some reading. I was wondering how many flat, dark and bias frames people take when doing astrophotography, and if people actually take bias frames or just incorporate them into the dark frames?

Cheers,
Matt.

[cometcatcher (Kevin)]

I only take dark frames and try and get at least 10. They say if you want to go very deep you need flats. I can understand that, but at this stage I only flat field in software.

[cfranks (Charles)]

I usually take 50 Flats for each filter, and at least 100 each of Darks and Bias frames. I take them during non-imaging weather, indoors, and just let it rip for the appropriate # of days. I live alone but that's not the reason.

Charles

[LightningNZ (Cam)]

lol, Charles!

There are very real benefits to having more correcting "exposure". Check out this thread here started by Shiraz (Ray): http://www.iceinspace.com.au/forum/s...d.php?t=122068

I take around 20-30 flats but I'm lucky enough to be able to do so with an LCD screen so my exposure times are very short. I'll likely increase the number of flats I take in future as there are very real benefits to having a very clean signal. Flats correct for dust or vignetting that is reducing the signal that you should be getting.

Bias shots are the shortest exposure you can get anyway, so you'll want lots of these - at least 50 I think. Too few of these can result in a smeared looking background if the sky shifts relative to the camera over time.

The value of darks depend on whether you take long exposures and whether your camera produces a lot of thermal noise. It's essential to take them at the temperature as the temp you used to take your light images. I generally don't bother with them. If I move to 5 minutes subs I'll consider it - and then I'd consider 20 to a fairly minimal number.

Hope this helps,
Cam

[jsmoraes (Jorge)]

If dark is to clean electronic noises and sensor fault, 1 can be better than 1000.

How you can know which area are with electronic noise ? Noise is noise: it is not a repetitive signal.

Bad pixel will be always in the same area. We can have, with time, an increase of bad pixels.

If dark must be with the same temperature, ISO, time of exposition ... many darks can be an issue. If you take expositon from 4 up to 10 minutes ... 50 or 100 dark will be the half, or much more, night period. When you finished to make all daks ... the night gone !

Bias is only to know the level zero of black to a specific ISO.

Flat ... the flats ... They are used against vignetting. Some people believe that they are useful against dust over sensor or lens. But they are not to clean them. They can hide them, if they are small.
My old Canon 350D has a large damage on the sensor. If I use flat, the black damaged area is changed by an white damaged area. They kill the photo in the same way.

And flat change if you change accessories. This type of file is very important if you want a profissioal photo. In normal use, they can help to reduce the vignetting ... only reduce.

Therefore:
dark: we can use mainly against bad pixels.

Bias: no problem, we can use old or new bias. The camera will not change its behavior so fast.

Flats: can help or not. Some times to use flats is worse. They are more complicated to produce every time you make photos or make change in your set of equipments and accessories

I use 5 to 10 of each type of file. And, always they are old files. I create new ones each 3 months or 6 months.

Actually... I'm not profissional !

[mathewb]

Thanks for the feedback everyone.

After still more reading, and with the advice from you guys I think i'll go with 10 or so flat and bias frames, and around 20 dark frames. I can build up from there if need be.

Apparently flat frames can be taken of the twilight sky, just before dusk. Not a lot of time to capture the images but it apparently works very well (the professional astronomers use twilight/dusk flat frames).

Cheers,
Matt.

[LightningNZ (Cam)]

I get hardly any vignetting to speak of but dust close on filters closer to the sensor is a problem because they are almost at the focal plane and so cast shadows (dust donuts) that will show up if you stretch an image enough.

The reason why you need more than 1 of each type of calibrating image is that when you take an image (even a calibrating one) there are multiple sources of noise in them - the main one being read noise from the amplifier(s). So to average out this noise you need to take multiple images and average them. Taking a single bias, single dark and single flat will likely leave your images looking like noisy trash because the read noise for each image is overwhelming any signal that would make them helpful.

[5ash (Philip)]

I used to do a lot of video astronomy using G star /Samsung ccd cameras and as a rule took at least as many darks as lights. I have carried this through to long exposures through DSLRs and CCD Astro cameras. So as a rule I do as many darks , flats and biases as my lights..
Philip

[Shiraz (Ray)]

I think that it depends Matt.

If you have vignetting and/or dust bunnies, you must use flats - but flats can introduce noise, so you need enough - suggest one flat per light.

If you have a cooled Sony chip and no vignetting or dust, you can get by quite well without any calibration.

If you have a cooled Kodak chip, you will need lots of good flats and darks to control fixed pattern and dark noise.

If you have a DSLR, at the very least you must have lots of darks to compensate for the dark current and the way it varies with time.

Bias is included in darks, but it is essential to measure it in isolation if you have any software that scales the darks to compensate for temperature or time.

If you do narrowband imaging, you will generally need to take lots of darks to clean up the background.

If you want to dig down to the very deepest levels, or you image under bright sky, you will need to pay particular attention to flats - and dithering would also be a good idea.

Overall, though, Philip's approach is nice and conservative - you won't go too far wrong if you take his advice.

[mathewb]

Hmm ..

Ok, I am using a DSLR for the imaging so I know there will be noise, a lot of noise actually (10 year old canon).. I also know that dark/image frames need to have the same exposure times. But hat about the flats? If I take 10 images at 20 seconds, and 10 images at 60 seconds then do I take 10 flats at 20s and 10 flats at 60s?

Cheers,
Matt.

[Shiraz (Ray)]

OK, then you need plenty of darks to handle the dark current, but you may be able to get by without flats at all, particularly if you dither your images.

If you do decide that you need flats (eg if you have vignetting or a bright sky), take them with a fairly bright source so that you can use short exposures and then take one flat for every light, regardless of what the light exposure is - this will be overkill, but flats are easy to get, so why not. Suggest that you roughly half fill the histogram when taking flats.

[Terry B]

If you want to do any science with your images like photometry then flats are essential.
For pretty pics they still make fixing gradients easier but are not as essential.