Subject:
|
Re: Very small triangle and a quad that thinks its a tri
|
Newsgroups:
|
lugnet.cad.dev
|
Date:
|
Fri, 14 Jan 2005 01:36:30 GMT
|
Viewed:
|
1682 times
|
| |
|
|
In lugnet.cad.dev, William Howard wrote:
> In lugnet.cad.dev, Lars C. Hassing wrote:
> > In lugnet.cad.dev, Ross Crawford wrote:
> > > In lugnet.cad.dev, William Howard wrote:
> > > > Whilst working out the non-right-angled cylinder intersection sub-parts for the
> > > > Boat Rigging pieces I have run into two related problems
> > > >
> > > > 1) I have a triangle A-B-C that is so tiny that L3P reports it as co-linear and
> > > > L3Lab doesn't render it (it looks OK in MLCad though)
> > > > 2) I have a large quad C-B-D-E that L3P reports B-C-E as co-linear and L3Lab
> > > > doesn't render (but again looks OK in MLCad)
> > > >
> > > > Now the triangle and the quad share the very short edge B-C but are not
> > > > co-planar so I can't replace them by any larger combinations. Also, omitting
> > > > A-B-C and making the quad into a tri leaves obvious gaps as the sub-part is
> > > > scaled when it is used.
> > > >
> > > > Should I
> > > > a) ignore these "errors" as they arn't really errors and assume that other
> > > > renders will render them OK (like MLCad does)
> > > > b) pre-scale the sub-part (which seems counter-intuitative as it then doesn't
> > > > follow the pattern of the right-angled cylinder intersection sub-parts)
> > >
> > > I'm guessing this is a limitation of the accuracy of the l3p -check, so probably
> > > ignorable.
> >
> > I think so too.
> > William, could you please e-mail me the actual coordinates of A, B, etc.,
> > and perhaps the whole file.
> I'll send you the whole file (its not complete yet), but the values for the tri
> are
> A (0.989107, -1, 0)
> B (0.997785, -0.98677, 0.06652)
> C (0.997785, -1, 0.06652)
That's a small triangle!
Side lengths: 0.0684 0.0671 0.0132
The reason it doesn't show in L3Lab is that you have color coded it in your file!
Back in 1998 PovRay would crash if it was given a triangle with collinear vertices
and a texture (e.g. Color15).
Therefore I added some code to skip collinear lines if color is not 16.
And since L3Lab and L3P share code, your colored triangle is skipped.
I have now tested it with PovRay 3.5 and it seems to handle collinear triangles well
- at least it don't break.
So, I have now deleted the color check. This means that no triangles will be skipped,
even collinear ones.
Also I have revised the collinear check to take size into account.
Now, the check is that the ratio of height / max_side_length must be < 0.001.
Thanks for the file. Your part now passes the collinear check.
(ratio of ABC above is 0.189833).
/Lars
|
|
Message has 1 Reply:
Message is in Reply To:
7 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
|
|
|
|
