Additive manufacturing (AM) provides enormous process- ing flexibility, enabling novel part geometries and opti- mized designs. Access to a local heat source further per- mits the potential for local microstructure control on the scale of individual melt pools, which can enable local con- trol of part properties. In order to design tailored process- ing strategies for target microstructures, models predict- ing the columnar-to-equiaxed transition must be extended to the high solidification velocities and complex thermal histories present in AM. Here, we combine 3D charac- terization with advanced modeling techniques to develop a more complete understanding of the solidification pro- cess and evolution of microstructure during electron beam melting (EBM) of Inconel 718. Full calibration of existing microstructure prediction models demonstrates the differ- ences between AM processes and more conventional weld- ing techniques, underlying the need for accurate deter- mination of key parameters that can only be measured directly in 3D. The ability to combine multisensor data in a consistent 3D framework via data fusion algorithms is essential to fully leverage these advanced characteriza- tion approaches. Thermal modeling provides insight on microstructure development within isolated solidification events and demonstrates the role of Marangoni effects on controlling solidification behavior.