The impact of crowding on the efficiency of the deblending algorithm. We show the number of useful deblended spectra as a function of the number of bright sources in the simulation. To obtain a measure of the crowding that does not depend on the specific simulation setup (seeing, size of the field of view) we have normalized the number of sources by the number of resolution elements. Two criteria were used to identify a usefully deblended spectrum: a magnitude offset <0.1mag (left) and an offset in radial velocity <2 km s^-1 (right) between the input spectrum and the recovered one. Each plotted datapoint corresponds to the analysis of one mock datacube. In both panels, a grey solid line indicates the optimal case where a useful spectrum could be deblended for every bright star in the field of view.