Abstract: The invention concers a method for preparing a synthetic image of a scene from a camera, said scene comprising a plurality of objects defined by elementary meshes stored in a database, said method including: (a) a step of defining, via a central processing unit, a plurality of rays exiting from said camera towards said observed scene; (b) a step of processing said plurality of rays including: a sub-step (b4) of assembling the rays into beams; (c) a step of propagating said beams including: (d) a first step of determining, for each of said propagated beams, a subset of all the meshing elements which are intersected by said propagated beam; then (C2) a second step of calculating, via a graphics processing unit and for each of said propagated beams, intersections between the meshing elements of said subset and the rays of the propagated beams; (d) a step of generating said image of the scene following the propagating step.