Toroidal computation of the plasma response to the n = 2 (n is the toroidal mode number) resonant magnetic perturbation field, based on an H-mode plasma in DIII-D, is carried out for the purpose of investigating the role of the ideal versus resistive plasma response models while scanning the plasma safety factor (q) at the edge. Both response models, implemented in the MARS-F code (Liu et al 2000 Phys. Plasmas7 3681), show significant amplification of the kink-peeling response in certain q-windows. A longstanding issue addressed in this work is the sensitivity of the q-window versus smoothing of the X-point geometry of the plasma separatrix. For this purpose, scan of the safety factor in 2D space (q95, qa) is carried out, where the q-value at 95% of the equilibrium poloidal flux, q95, is scanned by varying the total plasma current, whilst the edge safety factor qa is varied by gradually smoothing the X-point geometry at fixed total plasma current. Transition to the edge-peeling amplification domain is well described by simple analytic curves relating q95 and qa in the (q95, qa) space. These analytic curves not only help to quantify the sensitivity of the q95-window versus qa variation, but also establish the q95 window in the asymptotic limit of infinite qa (i.e. in the presence of true X-points). The 'resonant' q95 values (3.6, 4.05, 4.5) are found to roughly exhibit 1/n periodicity at infinite qa. Both ideal and resistive plasma response models, as well as the computed field at different pick-up locations, yield the same analytic curves describing transition to strong edge-peeling response. Detailed analysis of the computed plasma response and comparison with experiments are also performed.