[gregbradley]

I am finding flats are too often a problem for my image processing.

They become more vital with these large chip cameras. I am mainly using a 16803 chip.

I have been doing flats this way:

1. Typically a white T-shirt or white translucent foam plastic sheet, placed over the end of the scope.

2. Point towards the west area at about 20 degrees up.

3. Take flats with the scope in focus and cooled to usual light exposure temps.

4. Exposure time usually from 3 seconds minimum to maybe 40 seconds for Ha at 1x1. I shoot them at dusk when it is just about starting to get dark. If I shoot less than 3 seconds the mechanical shutter is seen in the image sometimes.

5. I do flats for each filter I use at the binning I am using when imaging.

6. I take a dark at the same exposure length and subtract that from the flat.

7. I usually take 3 flat exposures per filter. I move fast and get all of them done quickly as it is amazing how fast the light level drops. I find around 20.000 ADU gets good results. I have read where higher ADU gives lower noise but I find it overcompensates when used. Perhaps I am making a mistake in how I do them but thats what I find so I stick to arund 20,000 ADU exposure levels.

8. I then use CCDstack and use mean combine with the dark subtracted to create a master.

9. I shoot fresh flats on a regular basis and after a camera move.

10. I sometimes find these flats do not even out the image well and either overcompensate leaving outer dim corners too bright or not enough.

I also am finding that if I callibrate using bias subtraction as well as flats it gets a better result. But then I am also often using adaptive darks where there is not an exact match for the darks and the light exposures.

I have mainly a 10 minute -35 or -30C dark I use. In the recent heat my camera only went to -25C. Exact dark matches may be more important for best results here.

I am wondering if sky flats are the way to go. Also being more precise with the darks.

Any suggestions, experience with flats?

Greg.