I have a question on maximally unipotent monodromy point (or large complex structure limit) of the family of polarized K3 surfaces $(X,L)$. It is known that the moduli space of such pair is given by the quotient $\mathcal{D}/G$ where $\mathcal{D}$ is the period domain and $G$ is some arithmetic group. There is a standard compactification, called the Bailly-Borell compactification, of $\overline{\mathcal{D}/G}$. The boundary $\overline{\mathcal{D}/G}\setminus \mathcal{D}/G$ consists of points and curves.

My question is which boundary point or curve is a maximally unipotent monodromy point of the family of polarized K3 surfaces $(X,L)$? In some papers a maximally unipotent monodromy point is identified with a cusp ("point" above) of the moduli space $\overline{\mathcal{D}/G}$. Why is that? I think it is virtually impossible to compute monodromy around boundary.