A reoccurring theme in computational mechanics in recent years is the need to reduce the size of large discrete models. One type of such a reduction is spatial dimensional reduction, which one may perform in cases where the solution in some region of a high-dimensional (highD) computational domain, say two-dimensional (2D), behaves in a low-dimensional (lowD) way, say one-dimensional (1D). The motivation in constructing a mixed-dimensional (say 2D-1D) model comes from the fact that solving the problem in its highD form everywhere may require a very large computational effort. The idea is thus to partly reduce the spatial dimension of the problem, in order to obtain a hybrid model which is much more efficient. One field of application where mixed-dimensional coupling is of special interest is that of the dynamics of elastic structures. Typically, the LowD model consists of the slender parts of the structure that have rod- or beam- or plate- or shell-like behavior, and which constitute most of the structure volume, while the HighD parts are the small regions that have to be modeled as 3D elastic bodies. In the present work, we consider the 2D-1D, 3D-2D and 3D-1D coupling of models, to form a single hybrid model, for linear elastodynamics. We present some recent advances and new coupling methods. A special challenge is the attack of problems involving out-of-plane bending. For example, in the case where the highD model consists of 3D elastic elements, and the lowD model consists of 1D beam elements, there is a mismatch in the type of degrees of freedom (DOFs) between the two models, as the latter includes rotation DOFs and the former does not. Another aspect that makes this a difficult problem is the significant difference between the type of differential equations and finite elements (C0 vs. C1) used for each part of the problem. We will present several approaches for the coupling of such models.