Q: What is that thumbprint-like pattern in the background of
my image?
A: Believe it or not, it is not a varying level of background
but a varying level of the variance of the background. It is
purely the result of the geometric correction used to undistort
the science data. If you examine the raw image (the one with the
.d0h extension) you will not see that pattern. Likewise, if
you apply a small amount of smoothing (say a 3x3 boxcar) to the
image, it will disappear. This pattern arises from the pixels
of the geometrically corrected image drifting in and out of
phase with the input pixels of the distorted image. Because of
the way the geometric correction program resamples the image,
when the pixels are in phase results in a value in the input
pixel being effectively copied to the output pixel whereas when
it is out of phase, the output pixel is effectively the average
value of two or more pixels in the input image. All images of
the same format that use the same geometric correction file
will have the same pattern. If this pattern is a problem, the
user can write their own geometric correction program which
doesn't do this (but believe us, it will have some other
problem instead).
Q: Why do I see curvy edges in my calibrated data?
A: These correspond to edges of the raw data frame. They appear
curved because of the nature of the geometric distortion model
which is essentially extrapolated near the edge and hence very
inaccurate. Do not believe the position of anything within a
few pixels of the edge.
Q: Correct me if I'm wrong, but there didn't seem to be any
correction for the background in the image.
A: You are not wrong, there is no background subtraction. The
dark count rate for the FOC varies within its orbit and there is
no tool that reliably predicts what the dark count rate will be.
Given that, it is relatively useless to perform a background
subtraction.
