Shapes optimisation, Episode II

One of my main difficulty now is to "group" shapes in the scenegraph, because there are to many objects to manage. I can't group an entire building when modeled because of two reasons :


 - My scenegraph JMonkey Engine can't group objects with different textures.
 - As I said before, all shapes are manipulated by the scenegraph and can't be grouped before the rendering.

So I made an algorithm in my modeler, that looks for shapes with identical attributes and post-process like same textures, same rotation axes, same altitudes... and group them in real time during modelisation.

This method is very light and save a huge amont of memory, especialy with big buildings. But the code is not that simple and requires some more work to be implemented.

Here are some visuals of the grouping system in action.