Rendering Constructions with OpenGL
Motivation
The 3D wireframe geometry rendered in the Universe and Tracing views has been specifically designed for the particularities of Dynamic Geometry, in particular for the need of a fast, dynamic (re)calculation of a scene every time a construction is deformed in the view by the user. If that approach gives some satisfaction as a whole, there are nevertheless aspects of 3D rendering that are not supported because of their high-cost computation. One of these issues is the treatments of the parts of a construction that are "hidden" behind other parts of it. The lack of such mechanism in the Universe means that objects are likely to be "wrongly" represented, giving a false impression of their spatial location.
Several attempts have been made to address this issue (tailored Z-buffering, collision detection, ...) without any success (either because of their time-consuming computation or the negligible impact of the solution). The main - unsatisfactory - assumption remains that, by being able to deform the construction or rotate the point of view, users will be able to overcome the graphical ambiguities introduced by the limited rendering engine.
Using OpenGL as Rendering Engine
So what about using dedicated 3D rendering mechanism such as OpenGL?
These is the situation for the current compromise for an OpenGL-enhanced view in Calques 3D: using its engine for the display of the figure in a separate window but without the support of direct manipulation for construction or deformation, operations only supported in the Universe view. But as with the other views in Calques 3D, the OpenGL view is dynamically linked to the Universe: deformation of the figure, adding or deleting objects, ... are immediately propagated and rendered.
