I describe some of the computational challenges posed by handling photons in lattice QCD. In Euclidean space, amplitudes containing final-state photons are expressed in terms of hadronic correlation functions with a strong infrared weighting, which are hard to control. Secondly, including electromagnetic corrections in hadronic observables is mandatory once the percent level of precision is reached, and this leads to higher n-point functions that need to be computed efficiently. Thirdly, these corrections also lead to a renormalization of the QCD parameters. Two techniques to overcome these challenges will be described in more detail, the use of coordinate-space methods and choosing an ultraviolet cutoff for the photons that lies well below the lattice cutoff scale.