Question on mesh collider performance

Is it just a mesh collider period that puts greater impact on performance or is there leeway in how complex that mesh is?

I’m trying to set up enemy colliders for a FPS. Capsules and boxes work fine for arms, legs, hands, head, etc. but trying to get as close to accurate as possible for the torso just is gonna happen with boxes or capsules, even if I use multiples. The Unity capsule collider seems to have about 10 vertices, so if you make a primitive mesh with not many more vertices than that, will it still be a huge performance difference?

Capsule collider doesn’t actually have any vertices… it’s all parametric, that is it’s calculated mathematically based on the parameters radius and height.

The more triangles, the more costly a mesh collider is. Additionally, a mesh collider should always be convex, (for one thing, it won’t collide with other mesh colliders unless they are also convex, and in Unity 5 non-convex mesh colliders won’t be allowed at all.)

As for performance consideration… it’s better to try it first, and see, than to spend a lot of time worrying about whether it will matter.

“Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.” - Donald Knuth

My goodness, That quote is just a gem!! seems you just read my mind here.

Thanks a lot.