Having a tech support team for a shared code base is about the only thing management did right. You can't just hand someone millions of lines of code and expect that they'd know how to use it and fully modify all of its subsystems. If you read industry stuff you'll typically see gamedev mention that using Unreal does not reduce headcount. It typically requires more people than an in-house engine if you're working in the AAA space. Inevitably the engine will not be able to do what you want and you have to modify it and build tooling around the new features. This is much more difficult to do when you don't have experience with the engine.
Gears 4 had like 10 rendering programmers. :V (The rest of the
credits are kinda interesting considering the scope of the game)
Edit: Look at Ubisoft, Activision/Blizzard, Bethesda. They have multiple engines that their studios are familiar with and they're maintained separately. Each one has been tailored over time for the type of game. EA is the only one that's tried to make an internal Unreal Engine, and it looks like it's failing spectacularly.
Indeed. Even looking back at the history of UE3, we've got a slew of horror stories for games that went beyond the third person shooting. Folks should recall Too Human or even Mass Effect 1. It should not be a surprise just how level-based Mass Effect 2 and 3 turned out, and even just look at how the Citadel/civilian areas evolved. It's not hard to piss on Bioware for their infamous UI issues throughout the Mass Effect games.
(clearly, starting over from scratch with Frostbite wasn't ideal after half a decade of iteration).
Arkham is very much level by level with a dash of smoke & mirrors to make it seem open-world. That clearly changed over the years with a singular studio focused on expanding their branch of UE3 (things got borked for different reasons with the PC branch of Arkham Knight, but that's a different story).
Netherrealms did a metric ton of work to get their branch of the engine up to speed for a Vs. Fighter. I mean, just imagine. It's a level-by-level side-view of a limited scene and they had so many troubles with hitting 60fps, and then a bunch of other issues (still have issues with load times considering...). Japanese studios making use of UE4 for Vs. Fighters aren't really impressing either.
More recently, we might even look to how the popularity of PUBG spurred Epic to finally optimize UE4 for their own erm.... purposes. Clearly, the PUBG folks have a lot of work themselves to do to this day.
Ark - I don't think I need to voice much about this at this point, but folks have to kind of sit back and think a bit about the scope.
-------------------
Context.
--------------
Back when Bioware was finishing up DAII, there was that huge question about what to do engine-wise since they were just trucking along with a rather archaic engine. Granted, perhaps they should have pulled a Bethesda while not pissing on the modding community, which has its own benefits for consumer mindset/expectations.
Heckin, Team Ninja switched engines (finally) for DOA6, and that's still a learning curve for their artists currently while still being a bit of a technical... oddity. The Xtreme branch of the DOA engine is a funny one, but I digress.
----------
CryEngine? Well, good luck with the technical support/documentation, and so on.
/AlRant