Flats - how many

[luka]

A "simple" question, what is a good number of flats to take and why?

Wherever I read about it the flats get put in the same basket as the darks and the biases. For example, the DSS page (http://deepskystacker.free.fr/english/theory.htm) talks about all three at the same time and about needing up to a 100 (of each) for the best results.

Now, in the dark/bias images the dark/bias signal has a similar order of magnitude as the noise. Making lots of images will smooth out the noise and leave almost only the dark/bias signal. Flats on the other hand have a very large light component, much larger than the noise. So why do we need lots of flats?

Just trying to reduce time standing in the cold :cold:

Thanks in advance,
Luka

[multiweb]

I routinely do 80 flats per filter. They range from 3s to 20s each. The more the better. Less noise. I do the same amount of "dark" flats.

[Atmos]

That makes me feel pretty lazy, I usually do 16 or 25 flats and the same for flat darks and darks.
A number of years ago I was doing some research into what makes a good flat for scientific purposes and what came out of that was that the ideal is to hit 1,000,000 electrons.

[trent_julie]

Hi Colin,
Did you put that research into a paper that I can read? it would be really interesting!

Cheers,

Trent

[The_bluester]

I do 100 flats and 100 dark flats, the ASI294 is not the easierst camera in the word on calibration so I aim for the flats to be as high quality as I can, 100 averages the read noise very well.


I couldn't do it without a flat panel, I do my flats at ten seconds so sky flats would be a problem. The benefit of that is that I can reuse a master dark flat for a fairly long time as I know what my flat exposure time will be. I don't use bias frames, the ASI294 does not perform well with them due to peculiarities in it's setup.

[The_bluester]

I did see it on some ASI6200 darks I have seen, but you have to stretch the data so hard to see it that it is IMO not all that relevant, I have seen plenty of images from CCD cams showing at least as much (And often more) banding than changes form sub to sub.

I reckon the best approach is just to shoot lots of subs to average it out with the rest of the read noise.

[Atmos]

It was just a project for my course so nothing out there.
Others research was published in the age of high well depth CCD where 1,000,000 electrons was required for the 8-20 e- detectors being used with well depths averaging 100,000 e- or more. Getting 1,000,000 electrons with a KAF-16803 is easy because a 50% well depth is 55,000 e- so 18 exposures. Doing it with an ASI1600 is a lot more painful as you need 2,000 exposures at Unity Gain :lol: It has 25x less noise though so you can do a respectable amount of exposures.

The most important part of flat fielding though is good calibration. A bad dark flat will destroy your flat fielding no matter how many flats you have :)

[lazjen]

You feel lazy? Hrm, I've been doing 10 flats and 20 darks. :)

[billdan]

I haven't done any flats for 18 months now .

My QHY-12 (OSC) and Coma corrector are always clean so I never have dust bunnies.
I just use the Startools Wipe module to fix up any vignetting.

[Terry B]

I do much less. 3 or 5 flats. I suppose it depends on how important the noise level is. The frames are only a few seconds each so dark current is minimal compared to the signal. Mine are used for spectra rather than pictures. This means I'm not stretching the images much to get the very dim regions.

[gregbradley]

I do between 3 and 6 and they are dusk flats with a white cloth over the end of the scope. I have to move fast as the light fades and you don't have that much time to capture flats for 7 filters some of which have both 1x1 binning and 2x2 binning.

I tend to only do one set of 3 to 6 for RGB usually the red filter.

I don't see any downside.

Total exposure time is more important than a little bit of noise coming from your flats.

I also tend not to use bias for flats or if I do its a separate bias not a flat with a bias subtracted so I can not use it if I am not getting a good flat correction.

On my CCDs I go for around 20,000 ADU. That seems to work well.
I tried using 30,000 but they usually overcorrect. 17-20,000 ADU seems to match the lights nicely and correct the vignetting well and any dust bunnies.

I also prefer to clean my filters and CCD glass rather than rely on flats to remove dust bunnies because they don't always without fail do a good job of that.

Cleanliness is next to perfect imagingness!

Greg.

Greg.

[luka]

Thanks to everybody for replying. Got some food for thought.