We investigate the exercise of locally extracting the real and imaginary parts of the four twist-2 Compton form factors (CFFs) $$ \left\{\mathcal{H},\mathcal{E},\overset{\sim }{\mathcal{H}},\overset{\sim }{\mathcal{E}}\right\} $$
H
E
H
~
E
~
which arise in the deeply virtual Compton scattering (DVCS) process e + p → e + p + γ. Neglecting dynamical higher-twist contributions, we find that there are a sufficient number of DVCS observables and degrees of freedom to extract all 8 leading quantities model-independently, exploiting the azimuthal dependence of the absolute cross sections across all possible beam and target polarizations at a common kinematical point in {Q, t, xB, y}. As an example, for typical JLab lab-frame kinematics, we simplify the reduced DVCS observables to their dominant terms, providing a sufficient number of equations for local determination of the twist-2 CFFs. We demonstrate the feasibility using harmonic fitting to both cross sections and beam spin asymmetries with both real and pseudo-data.