Due to finite thermal conductivity and the heterogeneous microstructure of Ti6Al4V, the temperature distribution within large components during thermal processing is highly heterogeneous on both, the macroscale and the microscale. To compute a spatial distribution of stresses at the microscale, a microdomain partition is prerequisite. By analysing representative micrographs, characteristic grain shapes are determined which serve as validation of numerically generated realistic microdomain partitions utilising the technique of spatial tessellations. By generalising the standard Voronoï tessellation, a more sophisticated tessellation, the Johnson–Mehl tessellation is introduced to capture these characteristics appropriately. The Johnson–Mehl cells grow isotropically around the kernels which result from an inhomogeneous Poisson point process, replicating the underlying phase evolution mechanism during thermal processing. In order to capture the anisotropy of the microstructure caused by preceding forging, a geometrical morphing is applied subsequently to the computation of the spatial tessellation. Comparison of the basic features of both, the experimentally derived micrographs and the numerically derived ones, reveals a good qualitative agreement.