first off, thanks to georg for coding this script. it really works wonders on images from my canon 50d, which exhibits vertical banding.
i do have some questions about proper use of this script.
from experimentation (and reading the development thread), the script seems to work best when the images are gradient-free. this makes sense considering how it works. i think this means that mean you should not run the script on calibrated subexposures which contain gradients (as all of mine do). that would also be very time-consuming and i'm not sure it's possible to automate that (without writing another script that repeatedly calls georg's script. and in my case the new script would have to rotate the images first.). however, this method would seem to be the most direct solution to my problem, outlined below.
the other option is to apply the script to an integrated image which has gone through DBE. no problem... except that in my case, i usually have 3-4 nights worth of data, and my pointing is never exact enough from night to night. this means that the integrated stack will have several rotation angles, and as such the banding is superposed over itself at different angles. this makes the script's job impossible. i suppose that if you have enough nights with different rotation angles then the banding could be rejected as noise, but so far i still see the banding in the final integrated image.
one way to handle this would be to keep track of all the rotation angles in the stack, and then after stacking everything together, go in and apply the script, then rotate to the first angle, apply the script again, and so on. i have not tried this. not sure how well this would work, as the angled banding will have 'holes' cut through them from the prior applications of the script.
what i have been doing is not good either - i've been stacking each night's work, applying DBE, applying the script, and then registering and integrating together all the fixed stacks from each night. this leaves me with a banding-free final image, but it usually has some new, weird gradients which are the effect of imperfect DBE on the sub-stacks. also now i'm stacking stacks which do not have the same exposure time and thus have different SNR, but i suppose pixinsight is the right tool to be using in this situation due to the weighting by noise feature. but the main problem is, i have not been entirely successful at eliminating the gradients from the final stacks.
does anyone have more ideas on how to apply this script in this situation? thanks.
