Subject:
|
Re: Parts as volumes (instead of surfaces)
|
Newsgroups:
|
lugnet.cad.dev.org.ldraw
|
Date:
|
Tue, 9 Apr 2002 21:06:51 GMT
|
Viewed:
|
316 times
|
| |
|
|
In lugnet.cad.dev.org.ldraw, Steve Bliss writes:
> In lugnet.cad.dev.org.ldraw, Damien Guichard wrote:
>
> > Anyway, the remaining problem is that most (if not all, including Brick 2 x
> > 4) ldraw parts are not convex but concave. Is there any simply mean to
> > extend BFC with metacommands that split ldraw parts in convex subpart? Does
> > the topic deserve the effort? Any interest at all? Just an idea to see what
> > reply.
>
> Wow, good questions. We know that some collision detection is already
> practical - Eric Olsen demonstrated that over a year ago. But
> decomposing LDraw parts into convex objects, that would be a trick.
> Each external box 5 (or pair of box 5's) would have to be split into 5
> different subparts.
>
> What about antistud tubes? How could those be decomposed?
>
> Steve
Stud4 primitive could be convex-decomposed just as Ring4 primitive is today.
More generally, just as any LDraw surface is convex-decomposed with just
triangles and quads (3,4 points respectively), I guess any LDraw volume can
be convex-decomposed with just tetrahedrons and pyramids (4,5 points
respectively). Just as a quad replaces 2 triangles, a pyramid replaces 2
tetrahedrons. Only the BFC-ed base face (triangle or quad) has to be
displayed. The vertex is only for volume definition.
So a simple cube requires 6 pyramids but is not slower to display (6 quads,
3 BFC-ed) than current "box.dat" that already requires 6 quads (also 3
BFC-ed) without providing any collision capability.
Of course an "invisible" version of tetrahedron and pyramid should be
available so that totally inner volumes do not increase display time.
So an antistud tube segment requires 6 pyramids (3 are invisible) but is not
slower to display than a current antistud tube segment (3 quads). Also these
pyramids are simple to define because only one new point (same apex for 6
pyramids) is defined. Moreover this new point is not hard to find because
just any inner point is a candidate.
Is this legacy of James Jessiman or just illusion ?
Can you give reference to Eric Olsen solution?
Damien
PS:
1. Collision detection does not address the "building" problem. Solution to
the "building" problem is the LCD project at http://news.lugnet.com/cad/dev/lcd
2. May be apex plus BFC-ed base is redundant because the apex already
orientates the base (apex side is inside)
|
|
Message has 1 Reply:
 | | Re: Parts as volumes (instead of surfaces)
|
| (...) So stud4 would be decomposed into 16 trapezoidal prisms? That sounds like a fair amount of mark up. (...) Nod, true. But I'd rather decompose a 1x1 brick into 6 volumes (4 walls, 1 top and the stud) than 46 (one volume for each surface (...) (22 years ago, 10-Apr-02, to lugnet.cad.dev.org.ldraw)
|
Message is in Reply To:
| | Re: Parts as volumes (instead of surfaces)
|
| (...) Wow, good questions. We know that some collision detection is already practical - Eric Olsen demonstrated that over a year ago. But decomposing LDraw parts into convex objects, that would be a trick. Each external box 5 (or pair of box 5's) (...) (22 years ago, 8-Apr-02, to lugnet.cad.dev.org.ldraw)
|
11 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
