Instead of doing a proper 1:1 comparison, I moved on top a different dataset (because that's what one does to avoid a proper test ). When I finallt do a proper test I don't expect life-altering results. In theory, there should be some subtle improvement in the noise profile, especially color noise, and perhaps the stars might avoid any small debayering artifacts brought on by the algorithm trying to detect and "preserve" sharp boundaries.
With that said, here is my M1 OSC data from last fall. 498x90s (=12:45) processed at full resolution then cropped down to 1920x1280 still at full resolution. Not something I'd likely present in public, but just to see. I *think* I was able to use a lighter touch with noise reduction, but I can't prove that yet.
Bayer drizzle
Re: Bayer drizzle
Thanks! (I mostly meant that I probably would not try to share the 100% resolution, unbinned version - but such is at least technically justifiable with CFA drizzle where it is not with debayering)
As for star profiles - it does something different, but I have no idea what it means. SVD sampling choices are rather different between the two.
Drizzled: Debayered:
As for star profiles - it does something different, but I have no idea what it means. SVD sampling choices are rather different between the two.
Drizzled: Debayered: