Fix Version/s: None
ComCam data has patterns in the biases along the x-axis (perpendicular to parallel transfers). The fitted overscan is therefore not removing all of the signal it should. Reversing the application of the overscan and bias correction should allow these patterns to be removed by an initial bias pass, allowing the overscan correction to remove the row-by-row signals.
Robert Lupton added a comment -
Looks good (and it works at NCSA). One comment on isr.doBiasBeforeOverscan on PR
Robert Lupton added a comment - Looks good (and it works at NCSA). One comment on isr.doBiasBeforeOverscan on PR
This runs reasonably cleanly through `ci_cpp_gen2`, with altered `constructCalib.py`
And with `-c isr.doBiasBeforeOverscan=True` added to the `runIsr.py` call.
Repeating on comCam data also appears successful:
Jenkins is running: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/32465/pipeline