On my quest to make lamb out of mutton, I've embarked on making a decent flat set for my astro fest images. Last night I took the scope out and focused on Altair using the bahtinov mask and locked the focuser. This afternoon from 5 - 5:30 I went out and too a load of images with a t-shirt over the scope - and we had a thread recently about things neighbours get up to
So anyhow, I now have over 200 frames from which to make a flat set to go with my lights and darks (I appreciate it would be an ideal match because of the time differential). I programmed the camera to take sets of 25 shots which me whisking the scope around in between each frame so as to get seemingly random bits of sky. For each set, I had pointed the scope east and above the trees and set the exposure so that the histogram on the camera was in the 2nd quadrant (between 25 and 50%).
As a result if the fading light, the exposure length was increased as time went by. There is also some variation in the histogram of the images due to my "random" whisking often (maybe 25-30%) catching an area of the sky with a tree in it (there's a kit of those around here )
So now...I have all these images, how best to select a good set?
Random pic from set of he same exposure length
I've read a couple of pages that talk about median pixel values, so a bunch of images with similar values? (regardless of exposure length?)
I battled with t-shirts and skyflats, and met with increasing frustration. Then I bought one of Peter Miller's light boxes. It's brilliant!!!
It has a dimmer switch that allows me to tune it to give me 0.1 sec flats, which can then be corrected with a simple bias-frame subtraction without the need for flat darks. I take 100 on each filter - it's so fast! (and effective)
Yes, flats should be taken with the image train in precisely the same position as the subs, and ideally, the same orientation (to avoid dust motes from shifting as the scope changes direction).
Yeah thanks Terry, figured as much so at best I'm going to alter the effect of vignetting, not necessarily for the better, but certainly not accurate.
All part of the learning experience though, I'll have to spare it more thought going forwards. It's just too tempting to rotate the camera on a per subject basis for better framing so I guess in the future I should get a square or round CCD
With a light box those rotation dilemmas go away Dunk. You take your flats each time before you change targets (not that I do more than 1 target per night - I'm finding I like to give each beauty a LOT of individual attention )
Camera position may not matter if your FOV is symmetric (won't work if you have an OAG prism in the light path, etc.) Any dust bunnies will typically be very close to the sensor and quite possibly inside the camera. It's worth a try...
Dunk: you can't just point the scope anywhere and expect to get even illumination. From the ACP manual:
Quote:
AutoFlat.vbs acquires sky flats using the minimum gradient position in the sky. This position is on the solar circle near the zenith, offset in the anti-solar direction by 15 degrees. This is close enough for most uses, including precision photometry. For more information and the theory behind this selection, see The Flat Sky: Calibration and Background Uniformity in Wide-Field Astronomical Images, Chromey & Hasselbacher, PASP 108: 944-949, October 1996.
Probably pointing close to the antisolar point would be good enough if you're not doing precision photometry
I've only got a little time to lose to find out, Rick, so I'll give it a go anyhow, how bad can it be
I still need to cherry pick some frames from the 200+ I'll start with a set that look similar then and see what happens! I could really use some image analysis software
Also, is there a magic number for how many of these work best?
So how do the experienced of you do it? Don't touch the camera orientation at all and do flats in the morning?
In my blissful ignorance, I didn't notice anyone doing this, but I probably wouldn't recognise it if I saw it
So am I better off leaving out the flats until I refine my practical technique?
I usually do dusk and dawn flats every day I'm imaging. I also have a light box which I use sometimes, e.g. if dusk or dawn is clouded out, but the sky flats work better.
I think it's worth having a go now. If the flats don't work you haven't really lost anything and you may learn something.
If you want to check if camera rotation makes a difference, take a few flats at one orientation and then turn the camera by 90 degrees and take a few more. Make two master flats from the results and then calibrate one with the other. If the result is a nice clean, uniform field then camera angle doesn't matter.
So I've just thrown together a set of 23 flats and added them to DSS for one of my recent efforts and it's not bad, but it's not free of effects...notably along the bottom left edge
Oh well, nothing ventured, nothing gained! I know next time...
Definitely some improvements there, Dunk, even if not perfect.
Wrt the "magic" number, there is but it requires a lot of analysis to figure out. For my KAF-16803 camera, I need to collect about half a million electrons, which is about 12 flats at around 30,000 ADU. For a camera with shallower well depth it's likely to be somewhere between 20 to 50 flats. Get as many as you can...
I never change the orientation of my camera. It is much easier to guide if the axis of the camera line up with RA and Dec.
for photometry you need much more accurate flats but not as much for imaging. Experiment and see.
OK... so I've been tinkering with the orientation of the camera to see the effect...looking at the master flats in an image viewer, there's definitely some distinct movement of the shades, and a couple of round spots too.
I've restocked the image with the set of flats that - I think - are least intrusive, there is a wedge on the left of the image that is starting to manifest. Let me know what you think - obviously it's not a perfect situation - I'm just experimenting and learning, and gaining and appreciation of my mistakes
So I've been simulating with the equipment...setup the mount and scope, programmed synscan with the date/time the pictures were taken...obviously, I have the images so I have been approximating the orientation of the camera. The only wildcard is the reducer/flattener, since it's circular and can rotate all the way around...although the camera threads on the end.
Anyhow, will need to plugin the new data and see what happens!
Which flats are good flats
)
I'll start with a set that look similar then and see what happens! I could really use some image analysis software 
