Hierarchical visibility determination using visibility groups is now functional Each engine uses some system to determine which objects on the scene are visible to player, those should be rendered, and which one are not. It is very important to reject as much as possible unnecessary details, but it is also important not to spend much time doing it. Less time engine spends in rejecting not visible details, more time is left for what is actually visible to player - graphic details, physics, AI…

We decided to take one step further in exploiting hardware feature called "occlusion query". Basically, modern 3D graphics boards are able to report how many pixels of a given object ended up visible on screen (that is to say, not occluded by some other, previously rendered objects, like walls or doors).

That way our new engine can completely skip rendering of objects that are occluded, and that greatly improves rendering performance! Even more, instead of testing object one-by-one, we can now test the whole group of objects and skip them all in one go, if it turns out that the whole group is not visible. This is basically our VSD (visible-surface-determination) algorithm. And it works great!