Re: pfLOD / Intersection Traversal Anomaly

New Message Reply Date view Thread view Subject view Author view

Larry Lachman (larry++at++spiffy.paradigmsim.com)
Wed, 03 Nov 1999 19:22:35 -0500


Angus Dorbie wrote:

> Larry Lachman wrote:
> >
> > Attention: Performer Team
> >
> > I am using Performer 2.2.5, and want to report an anomaly with pfLOD
> > nodes and the intersection traversal. I discovered that if range 0 of
> > a pfLOD is set to anything greater than 0.0f, then the intersection
> > traversal stops at the pfLOD and does not visit any of the pfLOD's
> > children. According to the man page for pfNodeIsectSegs:
> >
> > "For pfLODs, the default is to traverse only the child that would
> > be active at range 0."
>
> Doesn't this mean that nothing is active at range zero, so the manual
> and behaviour is currently correct?

No, it does NOT mean that. The behavior may be correct, but the manual does
not document the different traversal behaviors on a pfLOD.

Define "active". Is the child active if it is rendered? My understanding is it
is active. Run the program I submitted. If nothing is active at range zero,
(i.e., pfLODRange( lod, 0, 1.0f );) then child zero, the esprit, should never be
rendered. However, it is rendered, which is why it was not immediately apparent
as to why the intersection traversal stopped at the pfLOD; cull and draw did not
stop at the pfLOD when the eyepoint was between range 0 and range 1, while
intersection did.

> > But this is not the case given the scenario described above. This
> > anomaly occurs regardless of the hardware the program is run on. This
> > is either a Performer bug, or behavior that you will probably want to
> > document in the pfNodeIsectSegs and pfLOD man pages.
> >
> > For your convenience, I have enclosed a succinct Performer program
> > that reproduces this anomaly.

--
_______________________________________________________________

Larry Lachman WWW: http://www.multigen-paradigm.com MultiGen-Paradigm, Inc. larry++at++paradigmsim.com 14900 Landmark, Suite 400 (972) 960-2301 ext 287 voice Dallas, Texas 75240 USA (972) 960-2303 fax

Angus Dorbie wrote:

Larry Lachman wrote:
>
> Attention:  Performer Team
>
> I am using Performer 2.2.5, and want to report an anomaly with pfLOD
> nodes and the intersection traversal.  I discovered that if range 0 of
> a pfLOD is set to anything greater than 0.0f, then the intersection
> traversal stops at the pfLOD and does not visit any of the pfLOD's
> children.  According to the man page for pfNodeIsectSegs:
>
>     "For pfLODs, the default is to traverse only the child that would
> be active at range 0."

Doesn't this mean that nothing is active at range zero, so the manual
and behaviour is currently correct?

No,  it does NOT mean that.  The behavior may be correct, but the manual does not document the different traversal behaviors on a pfLOD.

Define "active".  Is the child active if it is rendered?  My understanding is it is active.  Run the program I submitted.  If nothing is active at range zero, (i.e., pfLODRange( lod, 0, 1.0f );) then child zero, the esprit, should never be rendered.  However, it is rendered, which is why it was not immediately apparent as to why the intersection traversal stopped at the pfLOD; cull and draw did not stop at the pfLOD when the eyepoint was between range  0 and range 1, while intersection did.
 

> But this is not the case given the scenario described above.  This
> anomaly occurs regardless of the hardware the program is run on.  This
> is either a Performer bug, or behavior that you will probably want to
> document in the pfNodeIsectSegs and pfLOD man pages.
>
> For your convenience, I have enclosed a succinct Performer program
> that reproduces this anomaly.
-- 
_______________________________________________________________

Larry Lachman                   WWW: http://www.multigen-paradigm.com
MultiGen-Paradigm, Inc.         larry++at++paradigmsim.com           
14900 Landmark, Suite 400       (972) 960-2301 ext 287  voice           
Dallas, Texas 75240   USA       (972) 960-2303          fax
 

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Wed Nov 03 1999 - 15:15:36 PST

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.