When are conditional lines needed?

[Peter Blomberg]

Cylindrical primitives have conditional lines at their ends with control points such that they assume a linear continuation of the shape. The control points are visualized with the red, blue, and purple lines. This is perfect when the shape is continuing as assumed, e.g. at rounded corners of rectangular objects (yellow and green lines).

When trying to avoid rotating primitives 45 degrees, the solution tend to be placing two 1-8 segments back to back. The conditional lines in the middle are on top of each other, but have different control points (turquoise lines). Neither conditional line is accurate and reviewers may point out "missing cond line". The solution is to add a third conditional line with correct control points.

However, other reviewers see this "superfluous" conditional line and demand it being removed. What is the best way?


  cylpic.png (Size: 3.64 KB / Downloads: 61)

PS. My personal preference is to launch a series of cyli without conditional lines at the ends.
Edit: Added text about red and blue lines.

[Chris Böhnke]

Do you have an example part where this causes an issue?

Also I am not really sure I understand the right side of your image. What do the blue and red lines mean? You are only talking about the teal ones.

Typically the "overlapping" cond line of the 2 1-8 prims would be considered "complementary cond lines" and should be fine.

The teal lines should be correct if placed at 90° angles.

[Travis Cobbs]

Cylindrical primitives have conditional lines at their ends with control points such that they assume a linear continuation of the shape. This is perfect when the shape is continuing as assumed, e.g. at rounded corners of rectangular objects (yellow and green lines).

When trying to avoid rotating primitives 45 degrees, the solution tend to be placing two 1-8 segments back to back. The conditional lines in the middle are on top of each other, but have different control points (turquoise lines). Neither conditional line is accurate and reviewers may point out "missing cond line". The solution is to add a third conditional line with correct control points.

However, other reviewers see this "superfluous" conditional line and demand it being removed. What is the best way?



PS. My personal preference is to launch a series of cyli without conditional lines at the ends.

Saying that two cyl primitives next to each other have inaccurate conditional lines is wrong, and adding an extra "correct" one is also wrong. The current conditional lines are the way they are on purpose, and the fact that two can be put next to each other and still produce correct output is intentional. The two conditional lines together produce an identical visual result to the single conditional lines used in the middle of the primitives. While it is true that having two is slightly inefficient, the alternative is to either create a whole new set of cylinder primitives, or to not use cylinder primitives at all in this case. IMO, neither option is preferable to what is currently done.

Granted, I guess I am biased, since if a new set of primitives were created, I would have to add code to LDView in order to support them in its primitive substitution. But even ignoring that, I feel that the effort required to support a whole new set far outweighs the tiny inefficiency introduced by the current set.

It would appear that reviewers placing holds for this situation need to be educated, but that shouldn't be too difficult.

[Peter Blomberg]

Thanks Chris and Travis!

I now understand that 'complementary cond lines' when two cylindrical prims are back to back is ok. As long as the radius is the same and the shape is convex, each of the cond lines engage "too late" from one side, but accurately from the other, thus together producing the net result of a correct cond line. Works even when the radii are different. Great!
   

So the issue comes when one of the two surfaces is missing its complementary cond line. E.g. when a quad or triangle is used adjacent to a cyl prim. In this case, one needs to add either a complementary cond line or a cond line with accurate control points.

Now consider the case where one cyl prim is convex and the other concave. In this case, both cond lines engage too early. The cond line isn't needed at all if the radii are the same. Apparently, this scenario will need to be avoided.

  cond3.png (Size: 6.46 KB / Downloads: 21)

[Philippe Hurbain]

 Apparently, this scenario will need to be avoided.

Yeah, in this case it's best to have one less segment in one of the cyli and add a quad joining them (and possibly some complementary condlines if needed  )