The boundary theory for symmetric Markov processes studied by Z.-Q.Chen and Fukushima describes how stochastic processes reflect from the inner domain to the boundary and how they exit to the outer domain and how they return, in the sense of the Dirichlet form. In particular, by the one-point compactification of the state space and considering infinity as the boundary, one can construct Markov processes starting from infinity. In this talk, we will explain that the capacities and the equilibrium measures of Markov processes with the Dirichlet boundary condition defined in the inner domain can be understood through the hitting probabilities and the hitting distributions of Markov processes extended to the outer domain. As examples, classical capacities such as the Newtonian capacity and the logarithmic capacity, as well as the half-plane capacity used to describe the (chordal) Loewner equation and the Schramm-Loewner evolution are expressed by using the hitting probability of Brownian excursions starting from infinity.