Hello. I'm back with low poly tire
I'm working on it during 3 days. Hope you like it. Texture need some work and i'm going to test it in game.
Is that 24 sections? That is not lo-poly as such anymore then, more like medium. For tirewalls etc. it is way too much, at least without LODing heavily.. We are looking at something like this in hi-poly car, LOD1, ie other players car 10m away is this vertex density. The screen area is what matter; how much screen area will this object take in game? If it's 100m away, it's 4-8 triangles or just a billboard or drawing turned off.. From say, 50m away, you get to double digits and only when you are 0.5m from the this, we should actually start to worry about it's details (and here we have to get level design in, how likely we are to get that close etc..)
One thing to remember is to add collision meshes separately: one shall NOT use 240 vertices in a collision mesh for this kind of an object. Even more if it is physics object, able to move.... Good guideline is that Unity, Source, Unreal, CryEng etc will not accept over 255 vertices on physic object convex collision meshes (and i think gMotor2 has similar restrictions or at least guidelines.). So this is good guideline for all physics object, no matter what game we are working on. Also that convex, it is quite important as it decreases our collision detection and resolve by 20 times at some cases (torus vs pyramid, square peg in a round hole). Even so that when it's repeating object, like in SObject, the absolute best is to make that separately, after level design has put it in the right place. SObjects are the worst for collisions so i haven't used them as collision objects in years.. I wrap them manually with the Wall tool and just click "render" = off from that wall and collision = on.
If it's high speed collision, thing get even more precise and strict, we need to have cerain vertex density due to our frame-by-frame physics engine and interpolation; we "slide" from frame but we only detect collision inside frames.. So we can go 300km/h, 120Hz physics engine, i think it is closer to 2m against stationary object, we can sink that much in before we have first "collision detected", try to resolve that paradox when you are over halfway inside wall, your front sticking out the other side
So we increase vertex density a bit, say 2m panel size so that we have at least one edge on every sq m. Edge to edge is cheapest collision detection and easiest to resolve. Google GJK algrotirthm