The factorization of scattering amplitudes in the soft and collinear limit at next-to-leading-power (NLP) has been investigated over the years in a variety of approaches. Soft photons in particular provide a clean probe of NLP effects and they have recently attracted the interest of the experimental community. In this talk, after a general overview to the methods employed in the field of perturbative power corrections, I will concentrate on the soft photon brehmsstrahlung as a case-study. In particular, motivated by some recent claims about the consistency of the underlying soft-photon theorem, I will discuss different formulations of the theorem, showing that one version is particularly suitable for the numerical computation with publicly available tools.
