Which Better Optimiazation Way For Modular Meshes?

[Longrid]

I can’t understand in general which optimization method is the best for models.
In the second option, people say that you can separate the faces in the mesh so that there are no seams. Fewer seams, less feedback to draw. But then light faces are possible.
And if you leave only two faces, then this argues with the rule that the walls should have a wall thickness of 20/30 centimeters. And there will also be light faces.
The third option has too many seams, but there will be no problems with the above problems.
Who solves this problem? Can you suggest the optimal solution to this problem?
I would be grateful to everyone for answers and help!)

[ZacD]

So seams/hard edges double verts, it’s a minor cost even on low end hardware, unless your entire mesh is using hard edges for every tri. You don’t need to completely separate the faces, that’s what’s happening with the geometry behind the scenes anyway.

If you’re using Lumen, Mesh Distance Fields, or even often with lightmass and baked lighting, solid closed off meshes work better.

[Longrid]

Thanks you)