"streaked" integrated masters - poss bad integration settings?

cmarcus

Well-known member
Dec 8, 2014
79
2
As per the attached image my integrated masters (using weighted BPP) show up these streaks that are not in the original subs. the effect is worse in O and S.
My first thought is that I need to change the rejection settings for integration.
I attach a screen shot of an individual sub and the integrated master showing these streaks
TIA for any input
Clifford Marcus
 

Attachments

robyx

PixInsight Staff
Jul 17, 2019
398
138
44
Zurich
Hi @cmarcus,

this noise pattern is called "walking noise" and it's due to the coexistence of two factors:

1. a residual noise pattern is present in your calibrated subs
2. your subs follow a constant drift along a direction

the combination of the two generates this noise pattern which emerges in the final integrated image in the form of a "walking" noise (you clearly see that it follows a well-defined direction). It's very hard to handle it once you have it BUT your best friend to kill it is applying dithering on your subs. The effect of dithering is to randomly move this pattern in all directions such that it gets "averaged out" in the master light and won't show up.

So it's not a BPP or any other software process fault, the acquired data itself is affected by this problem.

Robyx
 

cmarcus

Well-known member
Dec 8, 2014
79
2
Hi Robyx, thanks for your answer, I thought we were dithering (the data is from some time ago) but I'll make sure we are properly dithering in the future! We have had a problem with constant drift so I can understand that being a factor.
Best regards

Clifford
 

robyx

PixInsight Staff
Jul 17, 2019
398
138
44
Zurich
Maybe your dithering distance is not big enough, anyway It could be that you reduce it by dithering more but for sure you need to fix the drift! :)
 

cmarcus

Well-known member
Dec 8, 2014
79
2
Thanks for that, we have fixed the drift, and in fact are using different optics, but I won't let a drift issue drift again (f you see what I mean) without correcting it. FWIW this is what I managed to get out of the data.
 

Attachments