Natural objects, such as bones and watermelons, often have a heterogeneous composition and complex internal structures. Material properties inside the object can change abruptly or gradually, and representing such changes digitally can be problematic. Attribute functions represent physical properties distribution in the volumetric object. Modelling complex attributes within a volume is a complex task. There are several approaches to modelling attributes, but distance functions have gained popularity for heterogeneous object modelling because, in addition to their usefulness, they lead to predictability and intuitiveness. In this thesis, we consider a unified framework for heterogeneous volume modelling, specifically using distance fields. In particular, we tackle various issues associated with them such as the interpolation of volumetric attributes through time for shape transformation and intuitive and predictable interpolation of attributes inside a shape. To achieve these results, we rely on smooth approximate distance fields and interior distances. This thesis deals with outstanding issues in heterogeneous object modelling, and more specifically in modelling functionally graded materials and structures using different types of distances and approximation thereof. We demonstrate the benefits of heterogeneous volume modelling using smooth approximate distance fields with various applications, such as adaptive microstructures, morphological shape generation, shape driven interpolation of material properties through time and shape conforming interpolation of properties. Distance based modelling of attributes allows us to have a better parametrization of the object volume and design gradient properties across an object. This becomes more important nowadays with the growing interest in rapid prototyping and digital fabrication of heterogeneous objects and can find practical applications in different industries.