I have mentioned UNIGINE cause I've heard that Dual Universe powered by that engine. It isn't free, which seems interesting, cause the best things aren't free. And best things and "widely advertised" and "common" are not necessarily the same
Surely the best engine in this case is that one that your developers have the most relevant experience in? Unless you want to add the challenge of learning a new workflow on top of actually building a game.
I second the other comment. The best engine is going to be the one your devs are familiar with. And if nobody on your team has any prior experience in any engine, that's a huge problem which makes the risk of failure much bigger.
The main advantage of using Unity is its out of the box support for multiple platforms, and garbage collected runtime, other than that it's slower, and graphics are on the ok-ish side (the best looking Unity game i've seen is Battletech), and i'm not sure about its open-world capabilities (Firewatch and The Long Dark world sections were rather tiny), possibly there's a way around it.
It seems that Unreal Engine is a better choice (Mass Effects, for example), but its pure C++ (with lua scripting), which means x3-4 as much of code, and trickier interoperability.
I dunno, tough choice. Good luck with your project!
I would really like to see some more recent tests as Unity before the 2018.x releases had a "ux/easy of use/developer first" approach but they did significant improvements on rendering quality and efficiency on the 2018.x versions.
Technically, they're wrapping a native library, so it's possible to create a benchmark where there's no performance gap whatsoever. The bottleneck is at the intersection of the runtime enviroment with the library, which makes comparison difficult.
Don't forget that Unreal has Blueprints. You could practically finish an entire game on blueprints alone! I recall that The Solus Project was almost entirely blueprint based.
