Sure they can...
They can download a
cityscape skycraper set and a
chaulky moonscape and some
palm tree props and write a rather procedural relationship system for populating a globe model with these assets. That's plenty doable. The Matrix City Sample has a ways more accurate scale and nuanced level of detail / augmented replication, and that was just one stretch of one city. (Matrix of course was also a major production by the best in the business to show off this tech, but it's off the same playbook of how to use UE5.) It's assets and it's algorithms, and much of this "one dude"s work came from the work of a lot of other people (including, of course, the basic UE5 development suite itself.)
You're still raging against developers not delivering anything this long into the gen which holds up to the promise of the
2020 UE5 Unveiling (Lumen in the Land of Nanite.) I get it, that shit was cool and the wait has been unbearable; little homebrew kitbash projects meanwhile have been shown playing with UE5, and that stuff has made it even more painful to see what all seems so doable yet isn't happening in mainline games. Can't these teams of 500-1000 developers just do
this?
...But realistically, these homebrewers aren't showing shit that developers don't already know. The assets are out there if you decide to go that route, the engine is available and its experimental tech is generally holding up better than even promised, and the basics are easier and cheaper than ever. But the hard part isn't just algorithmically dotting a map with content; the hard part is making the game. You need to diversify all the areas so that there's flow and style, so it doesn't just blend feel like one flat, bland surface. You need play systems, so that there's something to do in these open spaces. You of course need animations instead of the standard UE avatar, you need a camera system, you need lighting design beyond basic GI, you need customized props and landscapes and sounds and whatnot, you need to optimize it and fit it in an executeable that can be reasonably delivered without breaking a player's SSD bank... You need to do the real work.
There's a billion of these "Graphically INSANE UE5 demo" videos showing stuff people have been toying with... but look on Steam for any actual UE5 "games" and you still won't find much. There's
Low Light Combat and
Market of Light, "
UE5 Shooter Game", demos for Stray Souls and The Wickie,
Versus and
Space Station 51 are in Early Access, other stuff, some
Itch-IO shit (get ready for some PT clones...), a couple
hacks of Lyra, but nothing like even the deluge of pennycandy apps you'd expect from a free release of a huge technology breakthrough, and nothing that actually in playable state holds up to any level of expected wow-factor seen in these "INSANE" demos.
All these kids in their basements and attics who can make "INSANE" demos can't actually produce a game, insane or otherwise. Even with off-the-shelf assets and generally available game blueprints/templates and an engine that practically doesn't choke anymore on kitbashing, you still rarely find even simple gamejam-type "Let's just throw shit in to look cool while they float around rings or something" projects playable. These folks, they know how to pick out really nice wallpaper; they don't know how to architect the house. (Or rather, some do know how to do that, but that's the hard stuff that takes time, and they haven't had that kind of time and resources to do it for real.) Stuff's out there to mess with, but nothing has released to demonstrate anything more than that we're still just on the cusp of where things are going.
You should look under the bed next time, maybe the saw guy, a predator and chucky are waiting there for you.
