We present the second part of our detailed analysis of the hot subdwarf O (sdO) and spectroscopic standard star BD+28 • 4211, in which we focus on the optical spectrum. This target was selected to revisit the more general question of how reliably the atmospheric parameters of a very hot star can be inferred from optical spectroscopy alone. Given its status as a spectrophotometric standard, spectral data of exceptional quality are available for BD+28 • 4211. In the first part of our study, we determined the abundances of some 11 metals detected in the atmosphere of BD+28 • 4211 using UV and far-UV spectra of the star and corroborated the fundamental parameters estimated in past studies (T eff ∼ 82 000 K, log g ∼ 6.2, and solar N(He)/N(H)). In this work, we aim at rederiving these secured parameters on the sole basis of high-quality optical spectra. A first grid of non-LTE line-blanketed model atmospheres, including C, N, O, Mg, Si, S, Fe, and Ni with the abundances derived from the UV spectrum, does not give satisfactory results when we apply a standard simultaneous fitting procedure to the observed H and He lines of our optical spectra. The line profiles are not finely reproduced and the resulting effective temperatures, in particular, are too low by ∼10 000 K. We next investigate the probable cause of this failure, that is, the importance of missing opacity sources on the atmospheric stratification. We thus compare line profiles computed from models with artificially boosted metallicities, from solar abundances to 15× these values. We find that the structural effects saturate for a metallicity of ∼10× solar, and use this to compute a second full grid of models and synthetic spectra. This metal-enriched grid allows us to achieve significantly improved spectral fits with models having the expected parameters. As an a posteriori test, we compared the detailed profiles of several model spectral lines with high-resolution spectra culled from archived HIRES observations. The agreement between our synthetic lines and the observed lines is very convincing. Our test case thus reveals that there is still a need for models with enhanced metallicity for better estimating the atmospheric parameters of objects such as hot subdwarfs and hot white dwarfs if only optical spectra are available.