i think you'll likely end up wanting to break it down into reuseable pieces so that only things local to the player can be loaded as needed and also so you benefit from instancing, however since you already have complete models that are organized a certain way, I think it wouldn't be a bad idea to drop those in your game as is and start testing to see if there is actually a problem which motivates a lot more work.
Basically I would want to test and only break it down to the next smallest degree if there is a clear indication that too many things are being loaded. So right now you might have entire buildings as a single model, and maybe you find this causes some sort of issue, so the next step might be to break them into floors, and maybe you find this works fine so then you don't need to go to the trouble to break them down to walls and doors and windows. in testing don't just play in editor but make an actual build and test it on target hardware. You don't have to be very technical to notice big problems but in order to profile to identify what is causing big problems you probably want to pay for a bit of consultation from an expert who can walk you through things - you can find some intro videos to using profiler on youtube which can help you narrow things down a little better than guessing, but properly using such tools does depend on having fundamental knowledge of how the engine works which you can't gain from a few youtube tutorials
A lot depends not just on the genre, but also platform, your time and budget, project goals, team composition and experience... a lot of things, so what is typical or best for one situation might not be the best thing for your own situation. For example if you have budget of 10,000 dollars, the granular workflow that 100,000,000 budget studio uses might simply not work for you because of time and money constraints. It might be a matter of simply accepting worse quality or performance, or reducing scope.
My experience seems to suggest that a modular workflow where buildings are constructed from units as small as walls and windows and doors is the slowest means of production, however it is obviously paying back at a certain scale, and for teams of certain composition it is pretty much only way to feasibly collaborate. But for other teams it might simply be a big waste of energy. But figuring out if you are at such a scale or not is only something you can determine through testing.
Its always the same. Straighten what you want straight, locally relax the rest to follow that straight edge. You could pin it, or like i do, just exlcude it from your selection
so yeah testing with max 2025 they certainly changed some things.
For once, the Unfold3d method doesnt support it and breaks the mesh, neat.
Better yet, there used to be an option where you just let the relax by polygons run, it could overshoot but usually on a low setting you could just start it and stop it when it looked good. now this option is gone. i need to figure out where it went to and if it can be returned. because now you have to just click the button a bazillion times
it garbels up the fingernails for some reason. so those will need its extra treatment, but this isnt a bad start at all. personally i am not a fan of the thumb in the center but lets work with it anyways
note how it garbles up the shoulder, really nice. i am sure my mighty old max 2011 wouldnt do that
why it garbles up the fingernails is clear now. i would suggest to split them off
oof i didnt unwrap with max 2025 yet. thats definitely gonna be a bit of a struggle. texture display is blurry shit, that new pivot in freeform mode is... shit
cool stuff... grumble... will fix this
anyways. fingernails will need a bit of cleanup. lots of overlaps here and there, even split top and bottom gives weird results
i will ignore these
welp here we go. feels a bit like one arm behind my back but yeah didnt unwrap yet in newer max versions
personally i would move the seam from the back to the bottom on the arm by a few loops which would give a better loop for straightening. the thumb in the middle, not sure if i am a fan or not. works here i guess. some geo wobbles make some things a little harder than needed but nothing too bad overall