| |
In lugnet.cad.dev, Travis Cobbs writes:
> In lugnet.cad.dev, Ross Crawford writes:
> > And after looking at the discussion Don linked to, I'm wondering if a
> > similar thing could be done for type 5s on cylinder edges - make the
> > second test point on the plane where a rectangle would be if it was • butted up against the cylinder - ie at *half* the angle between two
> > adjacent cylinder rectangles. That way two adjacent cylinder segments
> > "co-operate" to draw the line in the same way.
>
> Yeah; I actually posted a response to that thread suggesting just this.
Wow! This is so cool! These cooperating type 5 control points give
us both first order continuity between the cylinder and the tangent
plane, AND 2nd order continuity between adjacent cylinder parts. All
with just a few small changes to a couple of primitives and ZERO
changes to the parts files that use the primitives. I just love it
when the math works out like this!
So who's gonna do the repairs? I count around 15-20 cylinder, cone,
and sphere primitives. Does this approach apply to any others?
> > I also realised that this will still cause problems if the adjacent
> > object connects on the same plane as the last rect in the cylinder (/
> > sphere) - the only possible solutions I can see for that is extra
> > primitives without type 5s, or inlining the code & removing them
> > manually. But I would think that's pretty rare with LEGO bricks,
> > anyway?
>
> Well, it's only a problem if you don't add another type 5 line to the edge.
> And I agree that it's probably rare.
I don't think this is a problem at all. Looking at the individual
planes in the cylinder is not in the spirit of the primitive, which
is supposed to approximate a cylinder. Remember, in many cases we
substitute a better curve for these rough approximations before
rendering. I think if the adjacent object is in the same plane as
the last rect in the aproximation you'd want to place a type 2
seam line there, because it's not gonna look like a smooth join if
you substitute a better cylinder.
Also it would either be a flat or concave join, which would never [1]
show a type 5 line anyhow.
Don
[1] OK, maybe if it were transparent, but I don't like to see too
many lines on a transparent surface.
|
|
Message is in Reply To:
 | | Re: Type 5 lines at the edge of primitives
|
| (...) Yeah; I actually posted a response to that thread suggesting just this. (...) Well, it's only a problem if you don't add another type 5 line to the edge. And I agree that it's probably rare. --Travis Cobbs (23 years ago, 6-May-02, to lugnet.cad.dev)
|
12 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
