Thanks for the great work! @TimoBolkart How is the sparse matrix of mtx to calculated in the BFM_to_FLAME_corr.npz?