Subject:
|
Re: Colinear Vertices
|
Newsgroups:
|
lugnet.cad
|
Date:
|
Tue, 9 Dec 2008 22:37:11 GMT
|
Viewed:
|
6282 times
|
| |
|
|
In lugnet.cad, Timothy Gould wrote:
> --snip--
>
> > Actually, LDView pretends that the three points are a triangle, then calculates
> > the surface normal for that triangle. If that normal comes out with a 0 length,
> > it declares the vertices to be co-linear. So its threshold is based on floating
> > point round-off while calculating the surface normal.
> >
> > --Travis
>
> I assume that you divide by the vector lengths first (as in normalise them
> before taking the cross-product) but if you don't then you should.
No, I just take the cross product, and if the length (squared) is 0, I give an
error. I don't see how normalizing the length of the two vectors that I feed
into the cross product would improve things much here. Sure, it might mean that
my detection rate would be slightly different, but given that I'm relying on
floating point precision in the first place, I don't see how it would make a big
difference.
--Travis
|
|
Message has 1 Reply:
 | | Re: Colinear Vertices
|
| (...) It depends on what your threshold is but taking the difference (as you do in a cross product) of two big numbers is less accurate (very much so sometimes) than taking the difference of two smaller numbers. Even without this issue it means that (...) (16 years ago, 9-Dec-08, to lugnet.cad)
|
Message is in Reply To:
| | Re: Colinear Vertices
|
| --snip-- (...) I assume that you divide by the vector lengths first (as in normalise them before taking the cross-product) but if you don't then you should. Tim (16 years ago, 9-Dec-08, to lugnet.cad)
|
39 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
|
|
|
|
