| | | | | ...actually the problem is not specific to PrimGen (btw, thanks Paul for this
great tool!) but appears when some rounded primitive meet a straight surface:
conditional lines at the edges of these primitives appear when they shouldn't.
See http://www.brickshelf.com/gallery/Philo/Misc/primgenbug1.png. The right side
shows a 1-4cone24 and the unwanted line, the left side is a 1-4cyli (official
primitive) that behaves correctly.
The origin of this issue is obvious here:
http://www.brickshelf.com/gallery/Philo/Misc/primgenbug3.png, the control point
of the cylinder matches the flat surface, while the cone control point matches
the next section of a cone.
Note that the problem is not specific to cones, it also happens with toruses
(http://www.brickshelf.com/gallery/Philo/Misc/primgenbug2.png). A 1-4cyli
created with PrimGen also shows this behavior.
Philo
| | | | | | | | | | | | |
| |
| In lugnet.cad, Philippe Hurbain wrote:
> ...actually the problem is not specific to PrimGen (btw, thanks Paul
> for this great tool!) but appears when some rounded primitive meet a
> straight surface: conditional lines at the edges of these primitives
> appear when they shouldn't.
>
> See http://www.brickshelf.com/gallery/Philo/Misc/primgenbug1.png. The
> right side shows a 1-4cone24 and the unwanted line, the left side is
> a 1-4cyli (official primitive) that behaves correctly.
>
> The origin of this issue is obvious here:
> http://www.brickshelf.com/gallery/Philo/Misc/primgenbug3.png, the
> control point of the cylinder matches the flat surface, while the
> cone control point matches the next section of a cone.
>
> Note that the problem is not specific to cones, it also happens with
> toruses (http://www.brickshelf.com/gallery/Philo/Misc/primgenbug2.png).
> A 1-4cyli created with PrimGen also shows this behavior.
I think a solution that mostly works was invented around this
time frame:
http://news.lugnet.com/cad/dat/parts/?n=3375
Instead of using cylinder primitives with an optional line at one end
that expects to be mated with a similar cylinder primitive, you use
optional lines at both ends that share the work by dividing the curve
angle in half. This puts the optional line control points on the
tangent plane, which allows you to mate a cylinder primitive with
another cylinder or with a quad in the tangent plane. But it doesn't
work for S curves made of two cylinder primitives mated in the
opposite direction.
I don't think this was ever implemented for cylinders, but it might
have made it into some of the sphere primitives. More info here,
with pictures and some discussion of the primitive generator.
http://news.lugnet.com/cad/dat/parts/primitives/?n=350&t=i&v=a
Don
| | | | | | | | | | | | | | | | |
| |
| > I think a solution that mostly works was invented around this
> time frame:
>
> http://news.lugnet.com/cad/dat/parts/?n=3375
>
> Instead of using cylinder primitives with an optional line at one end
> that expects to be mated with a similar cylinder primitive, you use
> optional lines at both ends that share the work by dividing the curve
> angle in half. This puts the optional line control points on the
> tangent plane, which allows you to mate a cylinder primitive with
> another cylinder or with a quad in the tangent plane.
Yes, that's the way cylinder primitives are done, and it seems to work very
well. I think other primitives could be improved that way.
> But it doesn't
> work for S curves made of two cylinder primitives mated in the
> opposite direction.
That case is not that bad - the angle where the middle line appears is divided
by two, so it is an improvement. Moreover the line is better accepted in that
case since the surface is locally tangent to the eye direction.
> I don't think this was ever implemented for cylinders, but it might
> have made it into some of the sphere primitives. More info here,
> with pictures and some discussion of the primitive generator.
>
> http://news.lugnet.com/cad/dat/parts/primitives/?n=350&t=i&v=a
This thread perfectly describes the problem. Unfortunately Paul didn't update
PrimGen...
Philo
| | | | | | | | | | | | | | | | |
| |
| In lugnet.cad, Philippe Hurbain wrote:
> > I think a solution that mostly works was invented around this
> > time frame:
> >
> > http://news.lugnet.com/cad/dat/parts/?n=3375
> >
> > Instead of using cylinder primitives with an optional line at one end
> > that expects to be mated with a similar cylinder primitive, you use
> > optional lines at both ends that share the work by dividing the curve
> > angle in half. This puts the optional line control points on the
> > tangent plane, which allows you to mate a cylinder primitive with
> > another cylinder or with a quad in the tangent plane.
> Yes, that's the way cylinder primitives are done, and it seems to work very
> well. I think other primitives could be improved that way.
>
> > But it doesn't
> > work for S curves made of two cylinder primitives mated in the
> > opposite direction.
>
> That case is not that bad - the angle where the middle line appears is divided
> by two, so it is an improvement. Moreover the line is better accepted in that
> case since the surface is locally tangent to the eye direction.
>
> > I don't think this was ever implemented for cylinders, but it might
> > have made it into some of the sphere primitives. More info here,
> > with pictures and some discussion of the primitive generator.
> >
> > http://news.lugnet.com/cad/dat/parts/primitives/?n=350&t=i&v=a
>
> This thread perfectly describes the problem. Unfortunately Paul didn't update
> PrimGen...
>
> Philo
I believe this was corrected back in November of 2003. There have been other
updates to PrimGen since then.
Look under options, 'original style cyli and cones'.
I do not recall a discussion where it was ever decided to change to the "newer"
configuration so I set the option 'on' at start. (it does not save this setting)
Cheers,
Paul Easter
http://www.geocities.com/pneaster/prim-gen.htm
| | | | | | | | | | | | | | | | | | Paul,
> > This thread perfectly describes the problem. Unfortunately Paul didn't update
> > PrimGen...
Shame on me for these words! I should wear better glasses.
> I believe this was corrected back in November of 2003. There have been other
> updates to PrimGen since then.
> Look under options, 'original style cyli and cones'.
> I do not recall a discussion where it was ever decided to change to the "newer"
> configuration so I set the option 'on' at start. (it does not save this setting)
Thanks a lot, Paul. Indeed it solves that issue. The official cylinder
primitives are now new style so I guess the option could be set to off now
(though most cone primitives have not been updated).
Now I have a request: could the new tangent style be applied to inside and
outside torus primitives? I noticed yesterday an ugly line appearing at the
junction between a torus and a cylinder...
Thanks anyway for this great tool!
Philo
| | | | | | | | | | | | | | | | | | In lugnet.cad, Paul Easter wrote:
> In lugnet.cad, Philippe Hurbain wrote:
> I believe this was corrected back in November of 2003. There have been other
> updates to PrimGen since then.
> Look under options, 'original style cyli and cones'.
> I do not recall a discussion where it was ever decided to change to the "newer"
> configuration so I set the option 'on' at start. (it does not save this setting)
>
> Cheers,
> Paul Easter
Paul!
I'm glad you're still around. ya know I'm still struggling with the issues
listed here:
http://news.lugnet.com/cad/dev/?n=9725
any chance for an update?
bye, w.
| | | | | | |