That platformer model seems very appealing... Unlike the other styles, it is rare today, even among indies. Plus, warioland games are among the best platformers out there, even to this day!

Differs from game to game - I don't know what's most common in my case but I've had game ideas based on mechanics, works of fiction or real-world concepts or just single pictures that flashed through my mind for one reason or another. What's really interesting (well, to me at least) is that I've come up with many game ideas while making or just listening to music. Of course sometimes I would get inspired by another game's gameplay and then perform a reskin in my mind, changing the setting and expanding the mechanics accordingly to that setting. And during my mobile phase my ideas usually came to me based on monetisation or UI concepts that would work well specifically on mobile. Anyway, I'm sure that there's not a single right way to do it and a brilliant game can originate in almost any way.

I also really adore the concept of games based on a simple theme as it's common in game jams. Working with such limitations and some really abstract theme can be insanely inspiring and make you come up with ideas you wouldn't possibly come up with otherwise.

Seems to me like a big challenge is the tradeoff between realism and excessivw complexity. I mean to feel real a game needs detail, but too much detail turns it into a too-complicated mess that isn't fun and doesn't progress enough per hour spent.
There's also a tradeoff with open world vs mission priority. By that I mean that if the game's plot is an important quest about saving the world or some such, it feels wrong to me to get pushed into time-consuming minor quests just to level up. The people who would populate a game world not realistically care much about sending the player on fetch quests or gather quests if their doom is at hand.

Dang, reminds me of a number of ideas I had for a trollish RPG I was thinking up a few years ago (still might make it one day). One of them was to give the player a cat that follows him around and keeps blocking his path in narrow corridors in dungeons. During combat the cat would just run away and hide.

The idea first. Always.
Then some "deeper thinking" on the mechanics involved.
Then the graphical aspects.
Then I write and sketch it all down in a "roadmap".

And then I bury that "roadmap" in my drawer, where it will never be looked at by anybody. :o)