I found it interesting that the article refers to manifoldness as 'the Vertex to Vertex Rule':
What this means is that each triangle of the mesh must share 2 vertices with its adjacent triangles. For example one triangles vertex cannot lie on the side of a triangle, it must be attached to another vertex of the triangle.
There's an image in the article that illustrates this. What do you think? Is this a clear/better explanation?
I suppose the issue is the different definitions of manifold. I've had some models that I can't do much of anything to get them to upload, and others go up with out a problem. and sometimes just using different scales has worked to make it acceptable.
Yes! Size is definitely an issue that can occur when you use 'meters' as the size unit. In some cases you may run into the limited internal accuracy of the software that you use. Vertices are then 'welded' together and this can result in non-manifold meshes. I always use millimeters as the size unit to avoid this.
I like your definition better, Bart ("each edge is shared by exactly 2 faces"). Note in their example (Figure 1), the object is not 2-manifold even though it satisfies their vertex to vertex rule. There are four edges that have only one associated face.
The technology used by 3D protoyping delivers commercial advantages for product departments, specifically in speed, cost, and availability of almost instant 3d models created from CAD software. Engineers and design technicians can use 3d printers during the conceptual stages of product design to produce prototypes of new products.