Does the amount of uvmap islands affect performance for game models?

Support Modeling
1

I’m wondering if you have a half sphere for example with 5000 tris total and I then uvmap it so that each face is uvmapped individually (just for comparison purposes here). So each face is basically an island. Would a game using such mesh uvmapped in such way run little slower than if the whole mesh was uvmapped as one island instead? Does dividing the big island into smaller once result in performance decrease?

In mesh if you have two quad faces connected via one edge you have 6 vertices ([|]). But if you separate the quads ([] and []) you get 8 vertices which basically means the game needs to draw 2 more vertices which means less fps if you extrapolate it to bigger numbers. Does the same apply for uvmaps as well?

2

i think in the BGE split UVs are split vertices. not really sure. cant speak for other engines though.

3

Split vertices is exactly what I was wondering. I’d imagine if it is true for bge then it is probably true for all game engines because instead of one uv location there are now at least two uv locations for each vertice.