The availability of very high spatial resolution sensors has for the past few years allowed a precise description of atmospheric scenes for remote sensing and surveillance applications. Clouds presence in the field of view is one of the key factors limiting the performances of these sensors for target detections. However, in order to develop such detection algorithms for images with a fine spatial resolution, a fast 3D radiative transfer tool dedicated to scene generation is necessary to obtain large number of realistic cloud scenes. Three-dimensional effects become more important when going to higher model resolution. For that purpose, fast solutions are needed since three-dimensional radiative transfer solvers are computationally far too expensive. Two different strategies are presented in this paper. On the one hand, an optimization of the explicit method Spherical Harmonic Discrete Ordinate Method (SHDOM) developed by Evans, K. F. (1998), associated with a fast image rendering solution. On the other hand, a fast approximation of 3D radiative transfer.