Oh boy, did I get that one wrong. I really wasn’t aware of this Shift-Freehand object. Now that I’ve been smacked across the face with a 3D polyline though, I see the exact same behavior reported by the OP. Sorry about that, all.
So exactly what is a 3D polyline good for? The thing I find most interesting about it is that it is actually thin in weight when standing alone in open space, something that can’t really be achieved with ordinary edges or curves. I wonder if there’s some way to make this thin line type more accessible, or to assign this entity type to lines drawn using inferencing–it would really fill a big void (in my opinion) in SU’s capabilities–namely, 3D annotation. To explain my point, I will excerpt here one of my own wishlist items, previously posted in the old forum:
Improved 3D Annotation Support. By “annotation,” I’m including all forms of human-readable information applied to the model, expressed in words, numbers, and symbols, and the various line types used to apply them. This includes dimensions and tolerances, item callouts, local notes, and things like weld symbols and surface texture symbols. Just as there’s no reason that a product definition must be a 2D drawing, there’s no reason that all these forms of annotation shouldn’t be applied directly to a 3D model to good effect. That is, there’s nothing about any of these forms of annotation that makes them applicable only to 2D drawings. As I’ve mentioned before, there are ANSI/ASME and ISO standards that define standard practices for the application of annotation to the 3D model.
I’m not asking for a complete ASME dimensioning package. I’m just asking for the basic ability for someone to construct these elements manually, which presently can’t be done in SU at all. Here are some requests all having generally to do with improving annotation:
a. Thin Graphic Line Type for Annotation. The big obstacle to full annotation is that there is presently no line type that can be used to construct leader, extension, dimension, center, and phantom lines, which are all supposed to be thin and visually subordinate to object lines. As it stands, any line drawn by itself in space comes out as a profile line, which is thick. This new line type should be sensitive to inference so it can be positioned accurately relative to geometry and text and so it can be built into dashed line patterns.
@tt_su, do you see any possibility of developing a line with the thickness of a 3D polyline and the ability to be controlled precisely with inferencing in the construction of 3D annotation?