The optimum utilization of solar energy requires effective harvesting of both the direct and diffuse components of ground-reaching radiation. Although solar beams are typically key contributors to the total irradiance under cloudless conditions, the diffuse component becomes important especially in regions where clear skies are not dominant. Even if the cloud cover and cloud microphysics are known, it is not an easy task to estimate the diffuse irradiance at arbitrarily oriented sloped surfaces. This situation arises from the extreme difficulty in solving the radiative transfer equation in such a heterogeneous environment. Models of Homogeneous Skies (MHS) with a set of discrete sky types are commonly applied in determining irradiance or illuminance. However, experimental data usually differ significantly from data predicted by MHS, for initially daylighting purposes where a simple sky luminance distribution is most useful. In this paper we show that the MHS error can be exceptionally large, especially for some cardinal directions. The use of MHS could also lead to misinterpretation of a momentary sky state if different sky types are required to accurately simulate irradiances on arbitrarily oriented surfaces. Nevertheless, use of multiple sky types at one moment in time has no theoretical basis and thus is generally unacceptable. Therefore, we have demonstrate that a new Unified Model of Radiance Patterns (UMRP) is more consistent with physical measurement and is generally applicable to all situations, including clear-, partly cloudy, and, overcast-sky conditions and, could accurately predict vertical irradiances or illuminances more accurately than MHS. UMRP implements heterogeneity of cloud systems and provides radiance or luminance distributions depending on the characteristic size of clouds, their albedo and their altitude. While the error in the irradiance computations is approximately 6% for UMRP, it can reach a value of 70% or even more in case of MHS. UMRP-based modeling of radiative fields is now simple using the UniSky Simulator, a heavily optimized multithreading application (publicly available at www.unisky.say.ski?lang=en&page=aplikacia). We expect the UMRP concept to be a new choice for modelers and PV systems engineers to predict solar energy more accurately under any meteorological conditions. (C) 2016 Elsevier Ltd. All rights reserved.