What is correct CFA pattern for Kodak KAI-4020CM ?

[johnpane]

I have a series of raw images from a SBIG ST-4000XCM, which uses the Kodak KAI-4020CM chip. I have not been able to find documentation of the CFA pattern in my online searches, and the Debayer process is unable to determine it automatically. I exhaustively tried all of the possible patterns and none match the debayered versions of these same images that were also provided to me, making me wonder what other processing was done when the source converted the images.

Thanks,
John

[pfile]

https://www.digchip.com/datasheets/parts/datasheet/596/KAI-4020-pdf.php

looks like BGGR?

rob

[sharkmelley]

https://www.digchip.com/datasheets/parts/datasheet/596/KAI-4020-pdf.php

looks like BGGR?

rob

It might depend on whether or not the FITS file is flipped 

Mark

[pfile]

yeah unfortunately that's always the case

[johnpane]

Thank you both. Column readout anomalies stretch down from the top in the FITS file. They are shown coming from the bottom in this reference: https://diffractionlimited.com/wp-content/uploads/2016/03/ST4000_manual_supp.pdf . Is that a clue whether it is flipped?

Page 6 of the document pfile referenced shows the lower left corner as pixel 1,1, implying GRBG if no flipping occurs beyond what is documented there.

[pfile]

Thank you both. Column defects stretch down from the top in the FITS file. Is that a clue whether it is flipped?

Page 6 of the referenced document shows the lower left corner as pixel 1,1, implying GRBG.

well the flipping of the fits file could happen in software regardless of what the hardware is like. the problem is that there's no way to indicate in the fits header whether the file is written as "bottom-up" or "up-bottom" (byte 0 is the lower left of the image or byte 0 is the upper left). the same file can be interpreted both ways by the fits reader in PI, which can be set to either mode.

sometimes the easiest way to do this is to take a picture of a real-life scene and fiddle with the settings until the colors make sense to the eye. could be easier said than done with a telescope though.

rob

[johnpane]

well the flipping of the fits file could happen in software regardless of what the hardware is like. the problem is that there's no way to indicate in the fits header whether the file is written as "bottom-up" or "up-bottom" (byte 0 is the lower left of the image or byte 0 is the upper left). the same file can be interpreted both ways by the fits reader in PI, which can be set to either mode.

sometimes the easiest way to do this is to take a picture of a real-life scene and fiddle with the settings until the colors make sense to the eye. could be easier said than done with a telescope though.

rob

Thank you Rob. Note, I edited my response above after you saw it.

I don't have access to the camera. The scene has some redder and bluer stars but differences are hard to discern in a single subframe.