After launching demo all engine disadvantages came up. Processor loading is not distributed well — say, we had 30-40% without any changes to scene, which in not a Good Thing. Also we had some too complicated optimization methods.

First of all (with thanks to your feedback) we decided to code perspective correction and to re-write “honest” Z-sorting with some new optimization tricks.

After all we managed to create new, 5th version of engine which will have different architecture (not changed since 2.0). It will suppurt multiply cameras, perspective, Z-sorting using BSP-trees. “Changes listeners system” is already implemented.

And sure, we’re also making engine-server integration.