In BedLeveling.cpp runZProbe we have this code:
#if Z_PROBE_Z_OFFSET_MODE == 1
distance += Printer::zBedOffset; // We measured including coating, so we
// need to add coating thickness!
#else
// homig will go Printer::zBedOffset in addition, so correct here as well
distance -= Printer::zBedOffset; // We measured including coating, so we
// need to subtract coating thickness!
#endif
Printer::zBedOffset is bed coating. So measunring distance to bed gets reduced by that value for at least G30 probes.
For G32 there is this extra rule:
#if Z_PROBE_Z_OFFSET_MODE == 1
currentZ -= Printer::zBedOffset;
#endif
so no change in that case. The computation is
float zall = Printer::runZProbe(false, false, 1, false);
...
if (s >= 1) {
float zMax = Printer::runZMaxProbe();
if (zMax == ILLEGAL_Z_PROBE) {
return false;
}
zall += zMax - ENDSTOP_Z_BACK_ON_HOME + plane.z(Printer::currentPosition[X_AXIS], Printer::currentPosition[Y_AXIS]) - plane.z(0, 0);
Printer::zLength = zall;
}
So from this I would say increasing coating to positive reduces z max/z length. This is for latest version - as there were many errrors regarding autoleveling in past, older versions might behave different.