Full ray-tracing maps of gravitational lensing, constructed from N-Body simulations, represent a fundamental tool to interpret present and future weak lensing data. However the limitation of computational resources and storage capabilities severely restrict the number of realizations that can be performed in order to accurately sample both the cosmic shear models and covariance matrices. In this paper we present a halo model formalism for weak gravitational lensing that alleviates these issues by producing weak-lensing mocks at a reduced computational cost. Our model takes as input the halo population within a desired light-cone and the linear power spectrum of the underlined cosmological model. We examine the contribution given by the presence of substructures within haloes to the cosmic shear power spectrum and quantify it to the percent level. Our method allows us to reconstruct high-resolution convergence maps, for any desired source redshifts, of light-cones that realistically trace the matter density distribution in the universe, account for masked area and sample selections. We compare our analysis on the same large scale structures constructed using ray-tracing techniques and find very good agreements both in the linear and non-linear regimes up to few percent levels. The accuracy and speed of our method demonstrate the potential of our halo model for weak lensing statistics and the possibility to generate a large sample of convergence maps for different cosmological models as needed for the analysis of large galaxy redshift surveys.