Calibration darks, flats and biases - the very basics

[BrendanC]

I know, I know, this has been discussed endlessly here and elsewhere, but I've read lots and watched lots and some of it I understood and some of it I didn't. A lot of it I didn't, actually, mainly because every time I think I get it, someone else contradicts what I think I've 'got'.

I've reduced it all to this, regardless of the number of lights I take, assuming I'm using a DSLR, with all the below steps at the same ISO value as the shoot:

ANY TIME
Bias - 50 frames with the cap on, exposure time=shortest

DURING SHOOT
Darks - 50 with the cap on, exposure=as per shoot
Dark Flats - 25 with the cap on, exposure=AV mode
Flats - 25 with the cap off, diffuse white frames eg t-shirt with light behind it or morning sun, exposure=AV mode

I know everyone has their own take on this (which is kind of my problem understanding it) but, reducing this to the very, very, very basics, as a starting point for a total noob with calibration... would this work?

Edited January 26, 2020 by BrendanC

[Frank the Troll]

• I've asked about this myself, and got lots of different answers,, it would be good to have a defacto guide

Edited January 26, 2020 by Frank the Troll

[BrendanC]

There are lots of links out there, all of which claim to be 'the definitive guide' but every time I read one, I notice that they've said something that makes no sense, or they're unclear about a certain specific part of it, or there's a huge debate that ensues around very obscure technicalities.

I just want to know whether this extraordinarily basic 'recipe' I've come up with, for my specific situation of having a DSLR, could work, as very, very, very, very basic starting point, as a general rule. without going into any specifics whatsoever.

Edited January 26, 2020 by BrendanC

[wimvb]

With a dslr:

• Bias frames with the lens covered (a plastic lens cap is not enough), same iso as lights. Shortest exposure time possible. The more the merrier, but at least 50.
• Darks with the lens covered (a plastic lens cap is not enough), same iso and exposure time as lights, and as close to the same temperature as possible. As many as possible, at least 20.
• Flats, at the same aperture as the lights, if you use a lens. Preferably also the same iso, but not necessarily. Note the iso and exposure time. 
• Flat darks, at the same settings as flat frames (iso, time) with the optics covered.

Bias and dark frames correct electronics in the camera, so they need the same electronics settings: exposure time, iso and temperature, as the light frames.

Flats correct optical issues (shadows from edges and dust), so they need to be taken at the same optical settings as the light frames, same aperture, camera orientation, etc.

Flat darks correct the electronical issues of flat frames, so need the same camera settings as flats, iso, temperature, exposure time.

If you use darks and flat darks at the very same settings as lights and flats, you don't need bias frames. Otoh, darks don't always work with non-cooled dslrs, and you replace them with bias frames. You can also replace flat darks with bias frames.

[BrendanC]

Thanks for this, really appreciate the feedback - and here we go again...

When you say "Otoh, darks don't always work with non-cooled dslrs, and you replace them with bias frames."

Does this mean that, given I'm using a non-cooled DSLR, I should forget about darks and use bias frames instead?

"You can also replace flat darks with bias frames."

Does this mean I can, or I should? And if I replace dark flats with bias too, then does this mean I don't need to bother with darks at all?

In which case 'the recipe' becomes:

ANY TIME
Bias - 50 frames, optics covered, exposure time=shortest

DURING SHOOT
Flats - 25 with the cap off, diffuse white frames eg t-shirt with light behind it or morning sun, exposure=AV mode

Yes? No? Somewhere in between?

 

Edited January 27, 2020 by BrendanC

[wimvb]

8 minutes ago, BrendanC said:

When you say "Otoh, darks don't always work with non-cooled dslrs, and you replace them with bias frames."

 For darks to work, their temperature needs to match the temperature of the light frames. Since you can't control that temperature, darks may not work. Some astrophotographers get good results using darks. Others, including me when I used a dslr, don't see any benefit with darks, and don't use them. There's only one way to find out: experiment. Integrate a stack with, and the same stack without darks. Then compare the result. If the image is the same or better without darks, you don't need them and you can leave them out of your workflow. Since flat darks have a shorter exposure time than ordinary darks, usually you don't need them. You can then just replace them with bias. The frame types you need are then: lights, bias and flats.

16 minutes ago, BrendanC said:

In which case 'the recipe' becomes:

ANY TIME
Bias - 50 frames, optics covered, exposure time=shortest

DURING SHOOT
Flats - 25 with the cap off, diffuse white frames eg t-shirt with light behind it or morning sun

Correct

16 minutes ago, BrendanC said:

exposure=AV mode

I always use manual settings for AP. That way I know the settings will be the same. In any case, make sure that iso is the same. 

[feilimb]

1 hour ago, BrendanC said:

 

DURING SHOOT
Flats - 25 with the cap off, diffuse white frames eg t-shirt with light behind it or morning sun, exposure=AV mode

 

Note: you do not need to take your flat exposures during the shoot (ie. the night of the light frame exposures).  After shooting a sequence of light frames, and at end of a session - you can simply carry in your telescope / lens with camera still attached, and ensure that you do not move the camera from it's orientation, and do not change the focus of the telescope / lens.  Then you can eg. go to bed, and return to shooting flat frames the following day (or even a eg. few days later) when you have time.

Edited January 27, 2020 by feilimb

[Pompey Monkey]

Bias:   Yes

Flats:  Yes

For a non-cooled DSLR, darks and dark flats are pretty much the proverbial "urinating into the prevailing airflow" for reducing noise. and can actually make things worse.

If you are guiding, then implement an aggressive dithering plan (min 12 pixels between subs) and use sigma pixel rejection in the stacking process.

If you are not dithering, then you really should. This is the most effective noise reduction method for DSLR imaging.

[BrendanC]

Brilliant, thanks all. It's very much clearer now. I actually tried using darks in DSS and it really didn't work out. I think this could be why - it's just not suitable for my situation. I might try again sometime, but I'll also experiment with just flats.

Nice to know that I can do the flats after the shoot too rather than sticking around in the cold and dark, Gollum-like.

I'll also think about whether or not to use AV focus. I just did it because I read somewhere that you should, but I can see it introduces something random into the process that should be under control.

Finally, @Pompey Monkey I'm not guiding, but when you say I should be dithering, do you mean I should be whether or not I'm guiding? In which case, is this something I should be looking at my camera to control (Canon EOS1000D), or software? I thought DSS could handle this sort of thing? I realise this isn't strictly related to calibration but I'd like to tie this final one off, now I'm a bit clearer on the other stuff.

[feilimb]

Guiding is required if you wish to shoot 'long' light frame exposures, and do not wish to have either star trails or egg shaped stars in the lights.  Guiding refers to the use of a smaller telescope of short focal length (eg. 100-200mm) mounted on the main telescope (although there are other ways to achieve it - via the use of an 'off axis guider').  With such a setup, the main telescope is mounted on an equatorial mount, and is 'tracking' the stars - but such tracking is never perfect, and it is the job of the guide scope (and associated software - which is not DSS) to 'nudge' the mount 'a little this way, and a little that way' when it needs to keep a star in the same location on the image frame.

If you shoot very short exposures, it is sometimes possible to avoid the need for guiding altogether.

I am assuming you are or will be using a telescope for imaging, but perhaps you are just using a camera with a lens of relatively short focal length (50-150mm) ?

[rickwayne]

Dithering, in case the term isn't clear, is displacing the optical train by a few pixels' worth every so often, so the same pattern of photons isn't falling on exactly the same spots on the sensor for frame after frame. That can lead to noise artifacts ("walking noise" is commonly called out).

It's frequently and conveniently used in conjunction with autoguiding because this is exactly what an autoguider does -- repoint the optics by a tiny bit (a fraction of a pixel) to compensate for mount imperfections or whatever during the exposure. The same software and hardware can be used between exposures to repoint by some number of pixels in a random direction, so that the resulting frames don't quite all line up and the noise artifacts are vanquished. Stacking software has to be able to align images anyway (frame-to-frame pointing is never perfect to begin with), so it's easy for it to deal with dithered images.

It can be done manually by changing where the optics are pointing by a little bit between frames, whether that be via a mount's hand controller (mildly tedious) or by changing a ballhead's position by a fraction (SUPER tedious).

[BrendanC]

Thanks all.

I was just wondering whether dithering - as in, changing the optical train, as you say - was something I should be doing given I'm not guiding and if so, whether that would be a hardware (camera/scope/mount) or software setting.

I'm going to leave dithering for another time. This was really about calibration and I'm happier with that now.