Flat field calibration

[mburge]

I didn't see a discussion on this, so thought I'd see what everyone thought. There is really nice support for dark libraries, thank you Aaron, what about a full calibration stack?

Easiest path would be a synthetic flat based off the camera location panel information. Lots of caveats here because of the typical lower quality of the lenses, but we could start with a radial approximation of the lens' geometry, taking into account any centering information. This would be a simple equation and of course would not take into account actual physical things like dust motes in the optical train. I do think it would provide a nice correction in edge illumination given the huge drop off on these fast fish eye lenses. To make it purely function based we could use a constant value for the bias (e.g., Siril style).

Of course a better method would be actually data driven with real flags, but what do you think of this as a start?

I see the nice python hook for generic processing if I wanted to try it out, but wondered if anyone had experience already? (And if Aaron was interested!)

[mburge]

Real flats, not flags...

[aaronwmorris]

I have thought about flat processing for all sky cameras, however, I do not think they are feasible. The math is well understood, so it is not an implementation problem.

There are a few complicating factors to using flats:

• A 0.1-0.2mm offset is very common. It does not sound like much but this usually translates to 50-100 pixels of offset in the X and Y axis. indi-allsky has settings for dealing with these offsets, so it is not impossible, but it is a complication.
• Flats are usually applied to the RAW image (bayered frame). Would it be possible to generate a synthetic bayered flat frame? If not, the flat would have to be applied after debayering. Not sure how well that would work.

I think only a synthetic flat would be in the realm of possibility... I do not think you could generate a real flat frame with an all sky camera. A traditional flat panel would not work. A dome shaped flat panel would have to be constructed to generate a real flat.

[mburge]

I agree, I think only a synthetic flat is practical. I did a rough version tonight on my mono 676 with zwo's 2.5mm f1.2 lens. Of course there is a 90% moon and I don't have any older imagery saved as FITS to process :-)

It did flatten the field nicely.

Let me think some more about your points and see what I can come up with.