Evolution with the number of iterations of the normalized ℓ₂ norm of the residuals (left) and normalized ℓ₁ norm of the diffuse and compact component for channel Ka and for various values for the modulus of β = γ. Reaching the upper limit of the data-fidelity constraint leads to a normalized residual ℓ₂ norm of 1. The ℓ₁ norm was normalized to the ℓ₁ norm of the input data in direct space (sum of the absolute values of the pixels in the frequency map). M corresponds to the highest modulus of the spherical harmonic coefficients in the input image.