One-Hot vs Dense Vectors

Dense vectors are compact and encode similarity; one-hot vectors do not.

Compared to one-hot vectors, what advantage do dense embedding vectors offer?