While working on a building, I had decided to remove the surfaces inside of rooms unseeable from windows or doorways at a distance from my model_1 but forgot that I was using it for my collisions which resulted in me going through walls and ceilings.
And so I wondered why not just use model_0 for the best and precise collision and what consequences would that have?
It really depends on how complex the mesh is. Mesh collider performance scales per vertex. Another thing to note is that while players do like precise collisions for gunplay, they often expect smoother experiences for movement and frames can sometimes get in the way of that.
For best performance you should be making everything out of primitives. If you value your time, a lower detail collision model is a common technique, and is the model_0 optimisation the game commonly does.
Comparing that my model_0 has about 1800 triangles in comparison to only 800 triangles for model_1 Il stick to model_1 for collision then. Guess my floor trims will then have to be visual and sound like rock when punched.
While being on the discussion of collisions what is the best way to effectively assign certain faces certain material sounds?
Ive started to use the mesh collider to assign the entire building a gravel_dynamic material for the walls and then to make the floors and doorframes sound wooden made a second collision that is slightly raised from the surface to make wood steps/sounds and also a third mesh collision for the floor that isnt wood I assigned concrete_dynamic so that it doesnt sound like walking on a boulder.
Unfortunately you’ll have to split the colliders and assign materials per collider.
Typically when I am modelling I’ll use an entirely separate model than model_0 for mesh collision for this purpose, I can split it up without worrying about the impact on looks or draw calls