I have for a while now been working on a game that uses tile based computer generated maps.
It uses static models arranged as tiles and rotated to proper angle.
However i have been wondering what might be the most performance efficient way to implement this.
I once thought of making the map one huge model, but seems collada doesnt like that.
Then i wondered if i could make larger pieces, such as a corridors as collada models, but then again many models will be used in the game.
Basically what i am asking is what the best tradeoff between model geometry and model instances performancewise is.