Is non-manifold bad?!


I’m having a bout of obsessive-compulsive disorder here. My question is: when is it okay to create separate meshes (i.e. non-manifold parts) of an object that would be connected in the real world?

For example, if I modelled a hammer and made separate objects for the head and shaft, did not merge them or connect the vertices and moved them together in 3D space then that would be fine right?

However, let’s just say I modelled a wheel nut spanner like this:

I merely intersected the two meshes in 3D space; I did not use any booleans or anything else to make the two pieces a solid object. It would look okay when rendered barring any Z-fighting. Intersecting them – even very cleanly – would create a lot more faces, and I would want to keep the poly count down.

So, is this an acceptable practice? If I wanted to turn pro at this, would I get away with it? Appreciate your thoughts, I am well aware that life is too short so please don’t bother pointing that out!


If you want different parts of a wheel spinner or hammer to blend together with subsurfing then you need to model them as a single mesh. Sometimes it’s okay to model different parts and just put them together if you don’t need smooth transition between the parts.

Continuing on from what CD said. This is really just a consideration of the number of polygons/level of detail you want to use on an object. I know what you mean in terms of having O.C.D. about the accuracy of the models you make and what not, but ultimately what your making is probably ether a still image, a video, or a piece of game content. (unless someone got blender to work with some kind of CAD like machining equipment) Basically, only do it better if you can tell it’s not right/accurate in the final product, otherwise over doing it is probably just a wasted effort.

Besides dude, if it’s for art, then it can be what ever the heck you want! There are people out there that take there kitchen garbage and sell it to a gallery after calling it art, if that isn’t a problem in today’s society I don’t not joining your meshes together will be ether. (ok, so that’s extreme example, but you get the point.)

unless someone got blender to work with some kind of CAD like machining equipment

As I understand it what you have here is not non-manifold, but two separate intersecting manifolds.

Intersecting geometry isn’t wrong in any practical artistic sense so long as you are aware of possible drawbacks (such as the z-fighting you mention), though I personally find it extremely morally offensive and would never knowingly or willingly engage in such depravity (except perhaps for low-spec real-time stuff, but even then only under duress).

what is meant by manifold

i tough that a non-manifold was an objecft wiht a concave faces

giving some problem with normals on faces that you have to manually flip

but what is this manifold thing - a new beast in blender?


Yes, that is a much better way to describe it. Two intersecting meshes acting as one object, with none of the vertices connected. Like making a crucifix by intersecting two separate cuboids.

Thanks for the replies everyone.

Hmm, I thought of a reason. What if I wanted to make the object semi-transparent? It would have to be a closed mesh or it would be obvious that I hadn’t joined the two halves of the object. Oh well…

Yes, if it were transparent it’d be better to make it a single mesh. It’s best to plan ahead so you can work with maximum efficiency. An opposite situation would be if you were making a high-poly mesh to bake to a normal map for a low-poly game model; in that case you shouldn’t care at all about closed meshes in the high-poly as it’d be completely unnecessary and a waste of time. So basically don’t worry about it unless you can think of specific reasons, like you did with the transparency scenario.