In this chapter we present some ideas and first results for efficient scene understanding in particular cases, where traditional visual understanding techniques are not sufficient or not possible. So, a technique is presented first, permitting to understand scenes containing mirrors, transparencies or shadows and, for this reason, difficult to understand with methods presented in previous chapters, computing a good viewpoint or exploring a scene. The presented technique uses object-space based contour drawing, in order to add apparent contour drawing, for real only objects of the scene, on rendered image. The second technique presented in this chapter faces the case where it is necessary, for understanding a scene, to have a look inside and outside the scene, but visual exploration is not possible, because the image of the scene must be printed. In such cases, the presented technique allows to add holes to the scene, in order to see simultaneously the outside of the scene and selected objects inside it.