We revisit the Slater half-occupation technique within the DFT-1/2 method to provide improved accuracy of 2D and 3D halide perovskites band gaps at a moderate computational cost. We propose an electron removal scheme from the halide states that drastically improve the predicted band gaps of 2D compounds. Concurrently, we compute the effective masses of the considered structures and show that DFT-1/2 describes them with a nice degree of accuracy when compared to available experimental data. Moreover, we assess the suitability of DFT-1/2 to compute the energy level alignments of this family of compounds. We compare the results in light of those we predict using the hybrid functional HSE06 and the many body perturbation theory within the $G0W0$ approximation. We discuss the limitations of our refined DFT-1/2 scheme, which tends to overestimate the downshift of the absolute valence energy levels of the layered perovskite systems. We anticipate that our results would be useful to initiate benchmark studies and investigate large systems such as heterostructures for which other approaches may be out of reach.