PROBLEM ANALYZE STARS AND INTEGRATION
Beta 126.96.36.199 Apple M Silicon 64-bits
OS Monteret 12.6.3 - MBAir with M2
When I try to integrate the images I get an error saying that some of the images could not be registered (5 of 7). This error disappears if I unselect the automatic star analysis, but then the result of the integration is very strange (see screenshot 1), as if it was trying to delete the stars from the final result. It gets worse the more images I add (see screenshot 2).
I have tested with the same images (7 sample images) in this old version and I don't get this problem, even with the automatic star analysis and the integration is correct (see screenshot 3).
I have compared the left side panels in both versions and everything is the same (except for the automatic star analysis, of course).
I don't know what could be happening...
I just read the thread posted this morning by @maxthebuilder, who seems to have had the same thing happen to him as me. I applied his solution (select all three options within the cosmetic correction section) and it worked for me as well. Thanks, Max, for pointing out the solution, I didn't know what else to do, I even uninstalled APP and reinstalled it again to see if I could find the key...
Anyway, you can close the thread and put a pin on Max's thread just in case it happens to more people...
PD. Perhaps those three options could be selected by default to avoid this problem?
Thanks for pointing to his workflow, that shouldn't be required though, so it may be interesting to have a look at your data as well. I'll forward your post to Mabula as well.
In the 2.0.0 beta's several things were improved in star analysis to make it more robust for more datasets. In your case, you do not need to enable cosmetic correction to make it work, allthough if it does that is fine. But in 3) Analyse Stars, you would need to increase the new cosmic ray/noise reducer a bit. The registration would fail if APP sees noise or cosmic rays as stars. By increasing the cosmic ray /noise reducer lower-limit area size to 6-8, probably all will work fine ;-). This is more efficient than increasing the cosmetic correction I think in terms of speed. In cosmetic correction, hot pixels would be removed (thus noise) so it makes it very likely that your registration issue was caused by noise in combination with small stars in your data as well.