Question on Flats

Folks - I have a question about flats.

Given that a flat is an image taken to determine the response of each pixel to light (which is therefore determining the effects on pixels of the optical train as opposed to the imaging chip), how close to your raw images focus position should the flats be taken?

For example, if you have dust motes in the optical train (who doesn't?), their size on an image will vary as focus varies.

So if I take my raw images at a focus of say 2000 (that's the distance from 'home' on my PDF focuser), in an ideal world, my flats should be taken at that same focus point. However, since focus changes throughout the night (as temperature changes), my focus point may start at 2000 but end at 2500 or even 3000.

Rather than take flats at a variety of focus positions, can I assume that for small focus changes (say plus/minus 500) that I can happily use a flat taken at the mid focus point?

e.g. for an imaging session:
Min focus point = 2200
Max focus point = 2700

Take flats at a focus point of 2450

In my case, one step on the PDF focuser is approx 0.004mm
Or another way, 1mm of focus travel is about 275 steps on the focuser.

Cheers
Dave

From what I've read focus does not have to be too critical for flats. Most of the carp that you are trying to get rid of using flats does not actually sit on the ccd chip which is what your prime focus is at. I've taken flats without even checking the focus of the SCT at the start of a nights run and they worked bang on.

Bigger factor is the exposure time. I usually go for an exposure which gives me an average background value of 2/3 full well capacity.

That's what I thought - I spent a fair bit of time getting my imager characterised (photon transfer curve) so I have the exposure times bang on based on full-well, linearity etc. ACP allows me to enter an ADU value and will take images that give me that value (+/- a bit).

As it happens, I do have a small number of dust motes that sit on the CCD chip but they don't seem to give me a problem. That said, the imager is probably due a service which I might get sorted out this year.

I suspect that when you say "carp", you're not talking about fish?

From what I can gather, the flats should be taken as close as possible to the original setup - focus position, camera rotation, filter position etc. If the focus position changes there is a very real chance that the illumination pattern will change, and the vignetting at the chip corners will differ from the data shots. This is likely to lead to odd gradients when processing.

If you have a small sensor without filters in a well collimated and clean scope then you may get away with differing flat conditions. I'd try though to make sure to keep the flats taken at the same time as the data shots for ease of processing.

I'm no expert, but this is the general impression I get from reading the CN and SGL threads on the subject.

Thanks Cathal - that again reinforces what I thought.

Taking flats at the same time as data shots is difficult though (and would add a lot of 'down-time' to the process).

I use a light box so typically, I do a run of data images through the night (usually unfiltered but occasionally with a BVRI filters) and note the the range of unfiltered focus positions. I then (at the next opportunity) run a session of flats - using the mid-point of the unfiltered focus position. When the session ruins, it automatically compensates for the different filters by using an offset file which contains my previosuly characterised filter offsets.

To do all that in the middle of an imaging run would be unproductive and a real pain...

I don't have a rotator nor does the position of the imager change so that remains constant and does not need to be accounted for.

The chip (Kodak 1001E) is pretty large (25mm square imaging area) and that's really where my original question came from.

That all said, with my current process flow, I've not seen any odd gradients when processing etc. So it sounds like the process I follow is robust enough for this particular setup.

So between yourself an Kieran, I'm happy that I don't need to change things at the moment.

Thanks for the feedback.

Dave

Dave,

As Kieran hit the nail one the head, the only thing i'd add is to stress the criticality of even field illumination. Uneven illumination is the biggest cause of gradients. I've been trying a few different techniques with varying degrees of success, once I got an optimal solution I'll write it up. The size of dust motes will be very unlikely to change by > 1 pixel due to temperature changes. Actually I'd suggest that airmass probably accounts for more of the focus change than temp anyway. Of course you need separate flats for each filter but you already knew that.

There is another argument regarding the typical ADU for flats, some say it should be close to mid-range others say a bit higher. I personally tend to stick to mid range. I'd be interested to hear views on this point.


D.