Cannot flat out vignetting

[johndias]

I followed the tutorial on Light Vortex for calibrating and registering my frames, but I cannot get the vignetting out of my image.  It's like the vignetting is reversed when all is done - maybe the flat calibration is over correcting?  I don't know, looking for advice here.

The setup is an Atik 383l+ with Astronomik 35mm LRGB filters.  The telescope is an Orion ED80T-CF and I have a Hotech FF in the image train.  The screen shot attached shows a subframe with flats calibration (left), the original raw frame and the master flat.

[pfile]

is the 383i+ a CMOS sensor? if so double check that the gain / offset of the calibration frames exactly match the flats and lights' gain / offset.

rob

[johndias]

Nope it's a KAF-8300

[pfile]

ok i was fooled once by an atik camera name that sounded like it had an 8300M but turned out to be cmos.

how are you calibrating the flats? i just use bias frames to calibrate my flats, the longest of which is about 5s. i run my 8300M at -30C and in 5s there's not really enough time for much signal to build up, so flat calibration with bias only works fine.

rob

[ngc1535]

Over/Under correction like the screenshots is often due to:
-double subtracted bias (or accidently subtracted a long dark)
- or very bright sky (typically strongly colored, not matching the spectrum of the flat field light source)

-adam

[johndias]

Flats were calibrated with master bias (actually a superbias) and master dark. 

Sounds like maybe I should try calibrating without the dark?

[pfile]

i guess it all depends on how your configured ImageCalibration - as adam notes if the dark is uncalibrated and you load a master bias and master dark into IC, then you've got a double bias subtraction unless you tell IC to calibrate the dark.

also if the dark was long-duration, then only calibrating it will probably also mess things up. you'd have to turn on dark optimization to resolve the dark signal mismatch between the dark and the flats... but then since the dark signal in the flat is pretty small, the scaling factor for the dark is going to be pretty extreme... and perhaps there's no point. so yeah, i guess i'd just try with the master bias alone and see how it goes.

rob

[johndias]

That did the trick!  Thanks Rob and Adam! 

[johndias]

Actually - turns out not only did I have to not calibrate the flats with master dark, but during integration, I had to disable normalization - any idea why that would cause the overcorrection?

[ngc1535]

John,

There are two kinds of normalization... one is for pixel rejection purposes and the other is the output method.
I assume you are talking about the output normalization (which is the first "Image Integration" panel).
So if I understand... you are creating a master Flat using Image Integration. If this is the case, you need to choose "Multiplicative" for the output normalization method. You didn't specify...and the default is additive with scaling. I suspect this could be a "gotcha" that many people fall into?  The additive part will definitely mess things up by adding/subtracting a pedestal to your flats during normalization... and it will give you the exact same problem as double subtracting a bias... which is also a pedestal type error. It would make sense that turning off the normalization method would work if you happen to be using a constant light source like a panel. It probably would not have worked for sky flats.

Can you confirm you did not use "Multiplicative" ? For the rejection normalization... Equalize Fluxes is typical... although I have a fairly constant light source (panel, no stars)...and in a 5 second exposure large outliers in terms of values are rare- I simply choose no rejection.

-adam

-*removed* (Ok Rob. lol)

[pfile]

come on man

[johndias]

Hey Adam,

So yes, I was speaking of output normalization, sorry I should have been specific.  I initially tried multiplicative (as the docs recommend) and then just to experiment used none, and my flats worked perfectly.

And yes, I'm using an Alnitak flip-flat so I expect that the light distribution to be even. 

So, I guess it makes sense given what you are saying, thanks!