Locality is a standard notion of finite model theory. There are two well known flavors of it, based on Hanf's and Gaifman's theorems. Essentially they say that structures that locally look alike cannot be distinguished by first-order sentences. Very recently these standard notions have been generalized in two ways. The first extension makes the notion of “looking alike” depend on logical indistinguishability, rather than isomorphism, of local neighborhoods. The second extension considers transformations defined by FO formulae, and requires that small neighborhoods be preserved by those transformations. In this survey we explain these new notions – as well as the standard ones – and show how they behave with respect to Hanf's and Gaifman's conditions.