Peter,
Screen stretches can be very deceiving with linear data. I'd suggest that you compare them statistically and ideally compare an old master bias with a new one.
I get quite a bit of bias drift on my cameras, even over a short period of time, so I use overscan calibration which accounts for the variation. My KAF-16803 also has a habit of developing new dodgy columns over time. Updating the Defect List when a new one pops up sorts that out.
Cheers,
Rick.
|