This is also the 80-20 rule in effect. The first 80% of a task takes only 20% of the effort, leading to an positive misperception of progress. However an 80% done program is useless (for other people, it may however fit a specific use case, which is most definitely not the purpose of a “game engine”). The remaining 20% of functionality takes 80% of the effort. This last 20% leads to a negative misperception of hitting a wall and being “stuck”.
Darn it I keep getting it all mixed up 
No. I advise you to stick to something viable to avoid a lot of frustration. Some people here told you that it isn’t doable for a single person and I totally agree with them.
I’d also advise, when looking at indie devs that have managed impressive projects, to take note of where they come from.
Some of the best indie games out there have been built by people with years of experience in the gaming industry, and in many cases, even one-man projects end up requiring a team.
Since we’re on the subject:
The number one rule is to NOT get ahead of yourself, under any circumstances. Don’t get cocky after you make a bit of progress, think about what you’ve learned and how it can be applied. So don’t go looking at projects made by tens to hundreds of people and think “oh, that’s easy”.
Indie game development is not a race, and the amount of time you take to develop a game is not necessarily related to the amount of dedication you have for the project. I could say I’m “dedicated” to building my engine, but I have other commitments that take priority, such as my last year of high school, and my job. It’s possible for me to plug every waking hour into my project, and while that may satisfy your criteria for dedication it’s also borderline stupid. It seems too easy to forget that games should be a result of passion, or enjoyment in what you do as a programmer, works of self-expression, and art. We all get to where we are because we learn from our progress, and from our mistakes, and whether the end-goal is money or just a game that you’d want to play, game development should be a satisfying, fun and enriching experience, not one where you sit at a desk and churn out code as fast as possible.
You may also want to consider how valuable “realism” is in games. Realism is a strange thing: the more realistic a game (or computer generated movie for that matter) becomes, the harder it gets to really immerse players / viewers. No doubt, GTA V is awesome, but as others said, that’s really mostly the work of awesome artists (and a huge budget). When you get down to it, the game mechanics are a whole different story. These are really great, but much closer to what one or a couple of smart dudes in a basement can achieve.
Well I really don’t have years of experience. I wanted to build something, So I read about the subject, Whenever I couldn’t figure something out I’d try to find a solution for it and I did. And I’m already using my engine for a game demo that I’m making, The only hurdles are things that I haven’t implemented and I’m not really interested in implementing them anyway because they can easily be added via script, The less pre-built features, The more dynamic an engine is because then it is not designed for a specific type of game.
Overall my answer was from my experience, What OP was asking is if it’s possible. And it most definitely is. You don’t need to have years of experience if you can learn fast.
Fast learning != effective learning != dedication
Dedication allows you to learn effectively. But it generally slows down the rate at which you learn.
Add that , imho , a perfect 3d Eva Green clone with enanced boobs riding nude a dragon
that doesn’t know where to go is worst than a decent “snake” clone 
Nude Dragon Rider is my new goal in life, with “the other kind of ride” DLC for extreme monetization. ;D
LOL
Yes yes, enough dropping hints…
Another interesting topic about this is Physically Based Materials
In my opinion raytracing in combination with radiosity has the biggest chances to appear realistic. But even that uses approximations.
But from an artistic point of view, realism isn’t very interesting. Reality, we have it all the time around us, and usually it’s quite bland and boring. Some painter once said, “A picture must be better than reality” - “better” leaves many ways of interpretation, but my point of view is, that “realism” isn’t a good goal, “impressive” is what you should aim for, and that can be done in a variety of styles.
I agree, if I wanted photorealism I would go outside and watch the world. The most impressive drawings I saw were these which, with their simplification exposed some of the nature of things, their inner concept.
Realism is pretty hard when you’ve only got 3 wavelengths of light to use to generate colours, too…
Cas
Think of the brids! They never see any sensible colors on our screens (since they have four color receptors, and not 3 like us …)
I’m still puzzled that there sometimes are two light mixtures which give us exactly the same color impression. We can’t tell yellow light from a mix of red and green light, because both trigger two of our receptors :o
Not sure if sarcastic
But as long as these 3 wavelengths match our eye’s 3 types of color-receptors, we’re all set for a healthy chunk of virtual realism
RGB colorspace, however, is indeed a mess, hardly useful for anything, except in the very last phase, on a monitor or image file.
Three channels are sufficient
this I know,
'cause the theory tells me so.
Infinite summations aren’t so hard
er…nothing’s coming to mind here.
Chromatic’s aren’t much of a problem. Luminance is. I’d like to see some the of HDR screens prototypes.