I think you're discounting the value of the tools Unity and UE offer to experienced developers/teams, and ignoring that outside a few exceptions, Indies haven't really been able to compete with AAA Devs since the PS2 era, when games more or less got programmed from scratch, or were adapted from the brittle code of previous titles.
Nearly every major studio or publisher has a similar toolset they've either built from scratch (RED Engine, Frostbite, Anvil, Decima, Id Tech, etc.) or license (like Unreal) and built on top of. Years of testing, R&D, and workflow refinement goes into making these toolchains extensible and useful for teams of all skill levels and functions, as well as to make them scale well to the needs of different titles.
The trade-off with these tools, is that their tremendous breadth can make working with them on complex projects their own knowledge domain for smaller teams, even as it abstracts away many of the complexities that come with developing your own engine.
If you're an engineering oriented developer who has the luxury of developing for a very restricted set of platforms and the time to debug their own tooling, with narrow, well-defined graphical requirements, a clear vision, and a technically inclined art team, then using a framework like Ogre makes perfect sense. Lightweight frameworks are a joy to work with, and you only have to add what you need on top of them to get the job done.
But iteration is slower, and you may spend months getting your tooling where it needs to be if you're going to work with a team.
Good luck onboarding new artists and game designers though. First you have to worry about training. After that, compatibility. Artists tend to have a workflow that works best for them, and even using open file formats, and established toolchains, they've got a gift for finding edge cases in your system. Your map editing toolchain also has to work for both the artists, and the designers.
Conversely, a mature engine like UE, or Unity has a wealth of crowdsourced documentation, and it's almost impossible to trip over an issue that someone else hasn't already triaged before you. New team members are almost guaranteed to know how to fulfill their responsibilities within the constraints of the engine's toolset, so they can get to iterating on prototypes much faster.
They're also typically extensible enough that the engineering guy(s) can put whatever efforts they would have contributed to designing a rendering engine, tools, and debugging platform issues into adding features unique to their title.
The featureset on these behemoths may be overwhelming, but it's more or less on par with what the 'pros' are using, so just by adopting one, you're virtually eliminating your technical capability gap with them. There is still a gap. With respect to tooling, Indies simply lack access and experience with parametric modelling tools like Houdini which greatly increase the efficiency of content-generation.
The rest of that gap can be broken down to experience, and manpower. Experience can be fixed, but few indies are able to throw the number of bodies at a project that someone like EA or UbiSoft can.
Engines allow anyone to make AAA level experiences with AAA levels of graphical fidelity now.
The output gap has become about art and content, something no indie can effectively compete with in terms of volume.
I agree that developing on large engines can cause you to hit a wall, and the engine essentially becomes the developer's world, but I think overall the proportion of people in the world who go further is the same, even if the proportion of people in the world cluelessly noodling with the low-cost space shuttle they've been given, and putting out garbage increases.
People incapable of competing have been allowed to join the market. But the democratization of engines has also given those with the potential to be great a much lower barrier of entry onto the development scene