Amount of noise reduction in superbias

[Diane Miller]

I'm using a Canon 7D Mk II.  My bias master looks pretty grainy but I'm not sure if that is what is considered noise that should be removed by a superbias.

I tried a superbias and got noticeably broken up vertical lines.  I had used 200 subs so used 6 for Multiscale layers.  Out of curiosity I ran it again using 7 and got a smother result.  Then I tried 8 and it was even smoother.

Which is best?  Is the parameter to be used dependent on the number of megapixels on the sensor, or some other characteristic of the camera?

Or would the difference be trivial when used to calibrate the darks and flats?  I'm not using darks (yet) and the flats look good -- to me -- when I apply STF.

[dnault42]

Superbias as it is works very well with CCD sensors.  It can be made to work with CMOS sensors as well but you have to go through a few additional steps as CMOS sensors have both horizontal and vertical banding due to how the nature of the read-out hardware.  Here's the gist of it:
1) Rename your master bias to 'bias'.
2) Run Superbias in column mode on your 'bias' image, rename the output 'superbias_v'
3) Apply PixelMath to the 'bias' image with this expression: bias - superbias_v + 0.1
4) Run Superbias in row mode on your 'bias' image, rename the output 'superbias_h'
5) Apply PixelMath to the 'bias' image with this expression: superbias_b + superbias_h - 0.1

Assuming PixelMath is set to defaults, outside of the expressions I specified, your 'bias' image should now contain a superbias image that has both the horizontal and vertical components of your master bias image.

Regards,
David