Hi Toha, yes a curve that was created as a result of intersections between surfaces can be pretty dense. But the thing that is most undesirable about the above surfaces is not that the control points are dense, it's that the surface shape is being affected by forcing it to hug the boundaries directly.
It's pretty subtle in this case but look at when I draw 2 lines connecting CVs of the triangular surface:
Then if I move the upper line down so it is on top of the lower one you can see they are not going in exactly the same direction:
It's a small amount but there's a slight twisting happening throughout that surface.
If the edges trace out the same silhouette from the side view, you would have better surface quality if the surface was a straight extrusion and the triangular border comes from trim curves on the surface rather than the surface itself hugging the border. That's what I'm trying to show by this:
Surfaces like that which are simple extended extrusions with trim boundaries on them will be generated when you do a boolean on a solid object cutting it with a 2D profile curve.
If you instead set up the model by building the 3D wireframe for it and then filling in areas using freeform surface commands between the edges you won't get trimmed surfaces, you'll get surfaces that hug directly to the edges. They can have a precise surface shaping in some cases like if the edges are the same exact shape or mirror images, but if they are of different lengths then the result will not be quite the same as an extrusion that has trims on it.
The main thing is that if you have extended shapes and use booleans with 2D curves as the cutting objects you will get the precise simple trimmed surfaces from that method automatically.
So when it's possible to do so it is better to use booleans to construct the model, especially using 2D curves as the cutting objects. This is where the NURBS modeling strategy is quite different from poly modeling.
- Michael
|