I'll put it this way: 6 months of hard-core coding in C++ won't get you as far as like a week of scripting in Unity.
That said, there is value in doing what's been done before and learning how everything works. I highly recommend the book Game Coding Complete, if you want a breakdown of a lot of what goes into making a game engine. If what you want to do is just build something for the sake of building it and learning how to do it, doing it all from scratch can be fun! Being able to say "I wrote every part of this" is really satisfying.
If your goal is to make a product that eventually gets released, you really ought to just use a game engine. If you want something to do for fun as a hobby, go crazy with learning all of the lower-level stuff that goes on in a game! Though, even if you want to build your own "engine," I'd recommend working with other ones a bit just to see how they work and to get some ideas on how you might structure things.
In any case, the scope of your project idea is way too big as-is for something you would want to do alone. The content alone would take a loooong time, let alone the programming. Start smaller and work your way up from there as you build your skills. Given enough time and practice, you could conceivably achieve some cool results, but obviously whatever you make won't be able to compete with something that had dozens of people with expert knowledge working for years. If nothing else, build small tech demos before tackling a big project; 3D is hard.
Anyway, the only way to really appreciate the scope of work that goes into existing game engines is to try and do it yourself. Play around with different existing ones, and play around with implementing some of the concepts yourself. You may find yourself thinking "man, this is absurdly difficult and boring, I'm just going to use Unity," or you may find yourself saying "wow, this is awesome! I love graphics programming!". The only way to know what you'll like is to try it out and see.