Strange Gradients and Patterns with BPP

[Francesco Wueest]

Hi all,
When using BPP I recognized that my calibrated lights have a gradient in it. Not all of them. It looks like that some of them are ok and some aren't. It doesnt matter if I use Masterflats or let the flats create by BPP. I always end up with gradients or strange patterns. If I do everything manually it just works perfect. I really have no clue what does BPP make to struggle. It becomes even more visible when I do integrate the whole calibrated BPP processed stack. Then the left side looks like overexposed.
In the attached picture you can see on the left side a calibrated light by BPP and on the right side the same light calibrated light manually. I have really no clue where this is coming from. It does happen also when I use for example only left pier side images and flats.

The failure is visible before staralignment. I made a Blink video so you see what I mean.You can clearly see the failures. All frames used in this video are from BPP after calibration. I just used the default BPP values. Has someone an idea where this is coming from?
https://drive.google.com/file/d/0B7X359X9fkz2cFBvQVJWV0JEc2c/view?usp=sharing

Direct comparison between BPP calibration and manual calibration using exactly the same files ( Masterdark, Masterflat, Superbias, light):
https://drive.google.com/file/d/0B7X359X9fkz2TGVERDVSRTNVQlU/view?usp=sharing

Francesco

[SteveP]

Hi Francesco

I'm getting a similar issue both with BPP and with the Image Calibration tool.  Perhaps understanding how the flats are applied would be good to know.
My issue began when I created some master flats in SII and Halpha and OIII using a QSI690.  I've noticed that the SII and Halpha flats have "diagonal" striations on them which, so far, I understand from reading various forums, is some kind of artefact of the chip visible at certain wavelengths.  However, assuming that the whole purpose of flats is to pick up this kind of thing and then eliminate it by applying to the subs after bias and dark subtraction, I was not too concerned at first.  I assumed that the BPP and IC tools would be somehow doing ( median(flat) * calibrated sub / flat ).  However the resulting images CLEARLY show the diagonal striations - almost as though they have been amplified rather then eliminated.  This is apparent when I use BPP and IC but not when I use PixelMath to apply the flat division.

Interested in anyone knows anything about this or can explain - maybe I should direct my/our question directly to the developers (who, by the way, have created a superb product in PI!!!!)

Steve