Testing Flats

[sreilly]

Is it possible to subtract a flat from a flat to see if there is any difference? I have taken 3-4 series of flats using my FlatMan XL panel and the rgb flats seem to do a good job at the taken PA but the luminance ones not so much. I need to remove the cover from the camera and checked the filters to be sure they are tight (SBIG STL-11002M and CW10) but I really don't see rotation in the flats. I'm using ACP to run the flats capture via script. As I said the rgb master flats (36 each for the PA and east/west) seem to work just fine. Of course the best test is the use on an image itself but I'm thinking there may be something else going on. It hasn't been a problem in the past but is on at least this one image. Images taken using MaxImDl 5.24 and calibrated and masters created using PI 1.8. Settings are as instructed in the tutorial on making master calibration frames here http://pixinsight.com/tutorials/master-frames/index.html

-Steve

[georg.viehoever]

I think you can do this after a LinearFit. This is necessary because your light sources have different spectral distribution, and exposure is likely different. After subtraction you should see only noise (more or less).
Georg

[sreilly]

Thanks Georg,

Does this mean taking the calibrated flats and combining them using LinearFit as the because as is I use Percentile Clipping? I've attached the setting screen shots of my flats calibration and flat integration settings I use for making my flats.

Thanks,

Steve

[georg.viehoever]

Steve.
what I meant was to use the LinearFit process to make the two images compaztible, then use PixelMath to subtract one from the other.
Georg

[pfile]

what if you just divided them instead? whatever the resultant value, as long as it is constant across the frame, your flats are the same.

rob

[georg.viehoever]

what if you just divided them instead? whatever the resultant value, as long as it is constant across the frame, your flats are the same....

yes, this would work as well, if you first substract the bias.
Georg