A phenomenological prediction for radiative neutron capture is presented and compared to recent compilations of Maxwellian average cross sections and average radiative widths. The basic parameters for it, photon strength functions and nuclear level densities near the neutron separation energy are extracted from data without an ad-hoc assumption about axial symmetry - at variance to common usage. A satisfactory description is reached with an astonishingly small number of global fit parameters when theoretical predictions on triaxiality are inserted into conventional calculations of radiative neutron capture. These predictions (constrained HFB calculations with the Gogny D1S interaction) were tabulated recently for a large number of nuclei. For the photon strength a parameterization of GDR shapes by the sum of three Lorentzians (TLO) is extrapolated to low energies. Level densities are influenced strongly by the significant collective enhancement based on the breaking of shape symmetry. In the predictions for both the replacement of axial symmetry by the less stringent requirement of invariance against rotation by 180{\deg} leads to a global set of parameters, which allows to cover the range in nuclear mass number A from 50 to 250. The impact of non-GDR modes adding to the low energy slope of photon strength is also discussed including recent data on photon scattering and other radiative processes. These are shown to be of minor importance for a comparison with experimental cross sections for neutron capture by even target nuclei. Here the triple Lorentzian (TLO) fit method for a parameterization of giant dipole resonances is normalized in accordance to the dipole sum rule and the droplet model with surface dissipation accounts well for positions and widths without additional parameters. Thus a reliable prediction also outside the valley of stability is expected.