Enter Siril, a stacking and processing tool that claims to provide background (gradient) extraction on a per-sub basis as part of the pre-processing before stacking. This is of GREAT interest to me, particularly for targets in the Northeast, which rise out of Hartford's light dome as the night progresses. Preliminary experimentation indicates that there is some possibility that Siril does what it says.
As an example I provide screenshots of the initial Autodev process in
First DeepSkyStacker using Kappa=3, Sigma=8. The presence of the gradient is pretty obvious: Second DeepSkyStacker AutoAdaptive Weighted Averaging, my first attempt to resolve the issue: I then moved on to Siril, starting with a standard stack: Finally a Siril stack with the per-sub background extraction applied before stacking: This one shows a bit more initial detail, and appears a bit noisier than the others. It also appears that the flats may have over-corrected a bit as the corners are rather too light. I am not surprised at this as the flats are exposed all the way to the right, which was needed by the first three stacks to work completely on the vignetting. I will shoot some shorter flats and stack with those to see if the over-correction can be fixed.
Another this I noticed about Siril is that it is more accurate reporting (and hopefully compensating for) field rotation over the course of the evening. I may also go back to my M33 data from a few years back to see if I can get any gradient relief there.
I am curious to hear if anyone else has tried Siril, and what your experiences with it have been. I will add additional info here as I continue trying it out.
Oh, I noticed the Siril stacks are mirror imaged vertically when opened in
Bob