Solar irradiance changes on a wide range of time scales and is a key driver of the Earth's climate where secular variability in particular is relevant. This is, however, not well understood and our knowledge relies on reconstructions based on sunspot numbers and similar proxies. The prime candidate to produce secular variability is a change in the surface coverage of small-scale magnetic elements. Direct observational determination of the flux emitted by these magnetic elements is difficult, especially as information covering a large spectral range is needed. Here we present a theoretical approach to this problem using intensity calculations from 3-D simulations of solar magneto-convection and compare these with the intensity calculations used in the successful semi-empirical S ATIRE models at disk centre. Eventually, such a comparison should lead to the removal of the last free parameter from S ATIRE-based irradiance reconstruction.