The issue is that in the FP3 the camera had 48MPs which were tied to a bayer array and could only output 12MPx.
In the FP4 each photosite can be processed separately hence the 48MPs gives 48MPx in selected situations. The output can be much smaller depending on a) user selection and b) what is offered by the software.
Of course I can’t really see any of the real detail, this is all hearsay.