Over the past two decades, the representation of the underground worlds has evolved considerably. Certainly, the classic cave topography, in plan, longitudinal section and cross section, remains the rule. But the arrival of the 3D has offered new ways of representing karstics networks, but also galleries or detailed objects (wall, stalagmite ...). The recent arrival of lasergrammetry and photogrammetry allows us to rethink, in depth, the underground images produced: we have gone from 2D to 3D. Yet we continue to visualize things in articles, books and computer screens. We project 3D views. We separate the galleries by floor. We separate the bottom and the ceiling, to offer 2D images, but now with high resolution and adapted to scientific issues. So, we go from 3D to 2D. This shows that 3D is not superior to 2D. It is only an additional option, a complementary means to study and visualize caves. Like all tools, it requires learning and appropriate use.