I checked the spectrace.fits files for SS96 comparing what we delivered to what they STScI has in the CRDS and there are differences. They set to zero what they assumed would be reference pixels. They also used a padding value definittion different from what we meant XPAD=20 for them means in the oversampled coordinate frame, we meant in native pixel equivalent. SO there is a mismatch in the size of the oversampled, padded image.
I checked the spectrace.fits files for SS96 comparing what we delivered to what they STScI has in the CRDS and there are differences. They set to zero what they assumed would be reference pixels. They also used a padding value definittion different from what we meant XPAD=20 for them means in the oversampled coordinate frame, we meant in native pixel equivalent. SO there is a mismatch in the size of the oversampled, padded image.