First: if 4m gives you the look you want, feel free to use that! It certainly gives more space for a third-person camera, for example.
Second: the point of modeling multiple divisions of the height looks something like this (very coarse):
You’ll note that these pieces are full height, 2/3 height, and 1/3 height.
If you use 4 meter size, you might want to make full height, 3/4 height, 1/2 height, and 1/4 height.
You will need to make slanted walls, and fitting/filling sections, that match up all the different combinations. You’ll also want to set your grid to match – although in this case, a grid of 100 will work fine and snap perfectly.
I don’t know about web sites that describe modular assets in detail, although I’m sure there are plenty – I learned about it back when Game Developer magazine still showed up on paper 