# What are the benefits of the different cylinder caps?

I've been modeling for quite some time but I’ve never bothered to ask this question.

There are several ways to subdivide a cylinder cap? I've been going across vertex to vertex (right example), but I’ve seen a great many models and more experienced modelers triangulate along the perimeter (left example). What are the benefits of triangulating the edges as opposed to cutting it across, triangulating to a center vertex or just leaving an N-Gon?

I suppose once the object is triangulated (in a game engine like Unity or Unreal), the way I’ve been doing it would lead to one extra polygon on the cap end. Is this the reason the crossing rectangles approach isn't used as much?

• I usually use a triangle fan. I doubt there is a "proper" way that works for all cases, it depends on where and how the model will be used.. (e.g. will it be subsurfed, will it be in a game, will it be animated/deformed.. etc.) – gandalf3 Apr 26 '15 at 1:44
• – user1853 Apr 26 '15 at 1:56
• I like the square approack because you can just press Ctrl+F > Grid Fill, but I don't know which is better – VRM Apr 26 '15 at 11:55
• I agree. It also makes selecting the edge loop a whole lot easier. – allen Apr 27 '15 at 0:12
• @Jerryno even in your own answer you use "i like", "I don't like" and "prefer" which are opinions. – zeffii Aug 12 '15 at 6:35

I like the topology clean whether it's game model or not - so here are all the caps you will ever need:

If you have more sides, you can downscale these caps into them - the corresponding sub-cap topologies are highlighted (4-side cap, 6-side cap, 8-side cap):

With this you can make very clean any size cap with grid-like topology (which is very good).

If the number of sides is too big, the marked quads on next example will be a problem. This is how to setup loops in such topology to keep it again mostly grid and fix the problem:

Not only for subdivision topologies it's always good to inset the cap like so (its good for smooth shading also, thus for normal-maps etc.):

I don't like the square approach for triangles, I would prefer the vertex-to-vertex quad approach if I had. Long thin quads are bad also but still better than triangles.

Creating a triangle fan could be only pardoned in mobile game models or when flat shaded. Long thin triangles and a vertex with more than 5 edges are both the worst from topology and shading standpoint.

No. Not without considering aspects of how the object will be used.

If you plan to subsurf the mesh, then you could help the shape by including an extra edge-loop slightly offset from the outer edge to help maintain the shape and shading. But on the inside it doesn't really matter, it becomes a matter of personal preference or convention, or engine limitation (these days not such a big deal)

• I haven't noticed a significant difference in shading regardless of how I subdivide the cap. I was curious, what would be 'proper' and why. I looked at professional projects, including ones built by Unity 5 and the Unreal 4 producers. They don't care too much about it. (the Unity 5 guys even stick to triangle fans and leave it at that). Mudbox gives a 'high-valence' error if such a model is imported into it though. I'm a bit confused about what that means and where such a thing wouldn't be allowed. For what i'm doing (simple game level design) it doesn't seem to make much difference. – allen Aug 12 '15 at 1:11
• yeah, it doesn't really matter how you fill immobile planar features (such as the bottom of a barrel), there is no single 'correct way'. Let how you intend to texture it decide your topology, if it helps align typography or some other feature.. Not a bad question, but highly subject to the individual situation. – zeffii Aug 12 '15 at 6:30