A simple procedure generating a multivariate density function that satisfies highasymmetry and polytonal dependency is defined and studied. Traditional approaches tomultivariate distributions develop functions that first establish joint density, withdependency inferred, and often poorly understood, as an adjunct at the end. The result isthat separate copula designs, such as the Clayton or skew-T, are sought to captureparticular dependency structures. The alternative approach presented here generates ajoint density function after the dependency is explicitly predefined. This approach, whichwe have termed the D distribution, has produced a variety of satisfactory results relatingto different univariate and multivariate distributions through the choice of an appropriatedependency transfer function. Here we show a new application of the transfer function toconstruct the flexible and hence highly applicable D distribution and D copula. Thisconcept is connected to copulas, neural networks, skew normal distributions andconditioning on hidden variables.