We compute nucleon and Roper electromagnetic elastic and transition form factors using a Poincaré-covariant, symmetry-preserving treatment of a vector × vector contact-interaction. Obtained thereby, the electromagnetic interactions of baryons are typically described by hard form factors. In contrasting this behaviour with that produced by a momentum-dependent interaction, one achieves comparisons which highlight that elastic scattering and resonance electroproduction experiments probe the evolution of the strong interaction's running masses and coupling to infrared momenta. For example, the existence, and location if so, of a zero in the ratio of nucleon Sachs form factors are strongly influenced by the running of the dressed-quark mass. In our description of the nucleon and its first excited state, diquark correlations are important. These composite and fully-interacting correlations are instrumental in producing a zero in the Dirac form factor of the proton's d-quark; and in determining the ratio of d-to-u valence-quark distributions at x = 1, as we show via a simple formula that expresses dv/uv(x = 1) in terms of the nucleon's diquark content. The contact interaction produces a first excitation of the nucleon that is constituted predominantly from axial-vector diquark correlations. This impacts greatly on the γ * p → P11(1440) form factors, our results for which are qualitatively in agreement with the trend of available data. Notably, our dressed-quark core contribution to F2 * (Q 2 ) exhibits a zero at Q 2 ≈ 0.5 m 2 N . Faddeev equation treatments of a hadron's dressed-quark core usually underestimate its magnetic properties, hence we consider the effect produced by a dressed-quark anomalous electromagnetic moment. Its inclusion much improves agreement with experiment. On the domain 0 < Q 2 2 GeV 2 , meson-cloud effects are conjectured to be important in making a realistic comparison between experiment and hadron structure calculations. We find that our computed helicity amplitudes are similar to the bare amplitudes inferred via coupled-channels analyses of the electroproduction process. This supports a view that extant hadron structure calculations, which typically omit meson-cloud effects, should directly be compared with the bare-masses, -couplings, etc., determined via coupled-channels analyses.