We extend the problem of finding an optimal structure with maximum load-bearing capacity to the case of multiple materials. We first consider a reinforcement optimization case where the structure consists of a fixed background matrix material with given strength properties and optimize the reinforcement topology within this material. We discuss the use of various isotropic and anisotropic strength criteria to model the reinforcing phase, including reinforcements with discrete orientations. In a second time, we investigate a bi-material formulation where we optimize the topology of two material phases simultaneously. Various choices for the material strength conditions are proposed and we apply this formulation to the optimization of pure tensile and compressive phases of a single material. In all cases, two optimization variants are proposed using concepts of convex optimization and limit analysis theory, namely maximizing the load-bearing capacity under a fixed volume constraint or minimizing the volume under a fixed loading. Both problems are convex and a penalization procedure is proposed. The underlying problems can be solved using conic programming solvers. Illustrative applications demonstrate the versatility of the proposed formulation, including the influence of the selected strength criteria, the possibility to obtain structures with members of fixed orientation or structures with different importance granted to tensile and compressive regions. Finally, we also draw a parallel with the generation of strut-and-tie models for the analysis of reinforced concrete structures.