20 Apr 2009 17:51
when experimenting with the 3DCoreSim, I observed the following: When switching the model to "wireframe cube" and rotating it automatically or by hand, some definitely invisible faces are not discarded properly. Then sometimes they ARE discarded.
Is that proper behaviour, or just a problem with the normals, or an actual bug? If it were a bug, this would effectively impact the performance since invisible faces would still render to the screen before being rendered over by frontmost pixels. When watching the wirframe sphere you can also see that there are definitely invisible faces not being discarded. The other models are too complex so I can't see it very well.
What do you think?