With all the tech being thrown around in this video are they making sense? Have they
hit the eureka moment or are they still rushing to the OMG moment? If Derek
or anyone could answer on the culling, AI, shader/occlusion rework if they are
making any progress? I'm not great at judging character and when Brian is talking
it seems he believes that everything is going great, I constantly question are they
all it on it or do the majority believe the are making great progress?
It's an interesting question about the technology, I would guess that if any of us knew enough about Cryengine to answer with confidence then we'd be working for CIG by now.
What follows is pure, half educated guess work: CIG do seem to be re-inventing the wheel a lot of the time. They like to throw a lot of technical keywords around which either have no well defined meaning (subsumption AI) or are comically misused (eg refactoring code = changing all the variable / method names). Whilst their solutions to the problems (eg volumetric fog, ambient occlusion) may produce a better result in the end, many of these things have an in-engine default implementation anyway. They seem to be wasting a lot of time and effort on relatively minor graphical upgrades.
Then there are the more crucial things such as AI and netcode where they could certainly benefit from a better implementation but seem to be taking their time doing it. AI Subsumption, to me, looks like the industry standard behavioural trees which they've just implemented in an external editor - clearly we don't have much more than a few screenshots to go on here. Their Network Bind Culling *should* reduce the amount of data being sent to each client and speed up the frame rates by ignoring updates which the player doesn't need. There should be industry standard ways to do this, the fact that they haven't done so yet presumably means that whatever solutions they've tried are making the server even slower - still, who knows.
