I’ve tried messing with mesh settings. It helps but not completely.
If I do showrendermesh then I can see the problem. I expect to see many quad facets running vertically down the vertical surfaces. This is what I see before I join the lower surface to the rest of the shape.
At the tip of the pointy part of the shape, the top and bottom flat surfaces are a fan of thin triangles radiating from a single point set back a fair distance from the tip.
Before I join the bottom surface - the triangle fans, top and bottom. are “joined” by lots of thin quads running down the vertical sides of the shape. (Technically they aren’t actually joined yet but, by eye, the quads that run down from the upper fan match up with the vertices of the lower triangle. This is the geometry that you would expect to see in the joined up object and would render well.)
But when I finally do the join - the mesh generating algorithm doesn’t join the two fans with quads - instead it seems to pick a vertical line at the very point of the shape then it joins the bases of the fan triangles to about 7, equally spaced, points on that line. This is what gives me the crinkled appearance.
The line that it picks also seems to be set inwards from the tip of the object and this is what gives the rendered shape its concave appearance.
I have found a dialog called, “Polygon Detailed Mesh Operations” and have played about with pretty much all of the options but nothing fixes the problem.
I’m going to try cutting the initial rail in two to see if I can increase the X axis symmetry of the model to see if this helps the mesh creating algorithm match vertices.Maybe it will then generate quads instead of triangle fans on the vertical surfaces.
I suppose that this isn’t really a bug, although it is very suboptimal. As I see it you have a NURBS representation of some joined surfaces and an algorithm has to convert that into polygons. There may not be any way of saying, “this surface is vertical and should be made of quads”. It’s obvious to a human but I don’t see a way in the UI to tell the software what I want.
Am I missing something?
Does the file format just store curves, surfaces, and control points or is there a way to say how the surfaces were generated and what the creators’ intentions were?