Hi mate,
It automatically points to the anti-solar point (point with least gradient) which is 90 degrees azimuth, and, 270 degrees altitude. I believe it turns tracking off -- I'll double-check the next time it does this.
In the past, I have followed Jason Jennings' method of just using luminance flats for all filters, but, because I've jumped on the automation bandwagon, I wish to be a bit more methodical about it, and, take flats for each filter. Furthermore, due to the crack in my sensor, it's now even more imperative to take flats for each filter, as the focus will affect the size of the crack. Last night/this morning, I only did flats for my hydrogen alpha filter, as that is what I was imaging with last night.
And, yes, you are able to record filter offsets for the flats. I have an issue with my RoboFocus/FocusMax at the moment, in that my centre point (best focus position) number keeps changing. There's no mechanical slippage that I can see, but, I suspect it has something to do with raising the duty cycle on the RoboFocus motor. It's currently at 0%; I've read that if I change it to 50%, it might help with this issue.
The camera was set to -20 degrees (CCD Commander sets the temperature, and waits for it to stabilise before proceeding).
I think you'll need to note down your filter offsets, using your luminance as your reference centre point. Then, just input them into CCD Commander, and, let it do the rest.
For now, once I've finished collecting all my data for each filter, I will manually go out and use my lightbox to collect flats for my current project (I'm loathe to play around with my RoboFocus settings while it is somewhat working at the moment -- I'm getting repeatable focus, so, will leave it as is). After this project is complete, I will use dusk/dawn flats.
H