I’m working on a game which targets mobile devices, currently for Iphone 4 and better. I’m currently working on a lowpoly missile which has 38 tris. I wonder if its better to use mesh collider for this shape or apply more or less 10 box colliders to fill the shape (as it has a pentagonal shape). How much diffrence is it in performance, maybe nothing to think about?
In your case mesh collider will be a better option… Also marking it as convex will be faster. 10 box collider will become heavier in your case as it will be equivalent to around 100 tris collider…
You can often get away with using a lot fewer collider primitives than you think. Look at all the clipping and “magic fences” you see in games you enjoy.
If, realistically, nothing is going through your tail fins, either let them stick out of the collider, or put a sphere covering 2/3rds.
I also suspect the standard box collider is using tricks to make it “count as” less than 6 faces, so maybe 10 box colliders = 30 faces. But – I’ve done this, but never for box vs. mesh – just make a scene that spawns more and more objects. Give them a tiny script to keep them in motion. See which one drops the frame rate faster.
Tested with both methods, no clear diffrence in my case, i guess i will go with mesh colliders as I need precision