When including spin-orbit coupling (SOC) in optical computations, I observed that optical conductivity and shift current calculated via the postw90.x program in Wannier90 v3.1.0 are nearly twice those computed without SOC, whereas other software packages do not exhibit this behavior. What causes this discrepancy?