As much as I would like to help you, and as much as I understand the good will being directed here, I can only say that I'm not a narrative writer. All I do in my spare time is some dumb novels and blogposts solely written for me. Game and script narrative is a fully dedicated profession with a very different approach, not just in the writing and construction, but also in the ways it is edited. I cannot comment on this. It is far beyond my scope and ability.
My only advice to you is the same advice you will hear everywhere else: just do it. Truthfully, sincerely, there's no other way to get better at writing. I can't give you a series of exercises to repeat over and over again, because unlike gym where we are building a specific set of gains, writing is fully dependent on your own personal taste. When I edit other people's work, I am fully conscious of their own writing style, their taste in pacing, their desire for hot steamy fuckery or cold insidious calculations (that may lead to hot steamy fuckery). I do not change any of those. Their personality has to shine through; their thoughts, their observations must be uniquely their own. If it doesn't, I tell them that you have nothing to say. Come back to me when you do.
You need to write and write until you find your own voice. It will be tempting to ape someone else's style. Sure, some people do that. I have heard of those who type out entire chapters to copy a specific voice. I will not begrudge you based on what you choose. But more than anything, I am just asking you to shove those words down onto the page. You will notice, soon enough, if you are a clean first drafter or if you're a vomit and edit tomorrow type of person. I am both, depending on my mood or scene.
The point is to find out who you are. Your ideal audience should be you.
I would also kindly dispel the ideal of a 'dream game' for it places a massive burden on yourself. When starting out, our taste is always all the way up there but our skill unfortunately lies at a different level. Take your time and be satisfied with your output. Your 'masterpiece' will come and go. You'll start work on it only to properly test it as vertical slice and find it trips on the first step. That is not failure. That is simply how it is. How you choose to go forward--to iterate or abandon--is up to you.
There will be another 'masterpiece' waiting. Ideas are cheap, after all.