Increased memory usage, poorer than expected performance, renderbin crashes (stopped when sharedgroups/links were removed), and geometry seemingly getting altered and mixed up (a leg on one of my characters suddenly replaced by a tree). That sort of thing. That was with 1.2 though... haven't tried with the most recent beta.
-----Original Message-----
From: Discussion list for Java 3D API
[mailto:[EMAIL PROTECTED]]On Behalf Of Lan Wu-Cavener
Sent: Friday, July 05, 2002 1:52 PM
To: [EMAIL PROTECTED]
Subject: Re: [JAVA3D] Trouble with Link/SharedGroup - is it a Java3D
bug??
I don't know what kind of inconsistency you had. In my application the
usage of Link and SharedGroup does work, but the frame rate is really bad
while navigating the scene. I also use the combination to distribute trees
over terrain.
At 08:13 AM 7/5/2002 -0500, you wrote:
>I had some inconsistencies with Links and shared groups as well, and even
>stopped using them when I couldn't figure out what I was doing wrong (memory
>use extreme, visual glitches, no perceived benefit to using them...) Will
>try to find my notes.
>
>Scott
>----- Original Message -----
>From: "Joerg 'Herkules' Plewe" <[EMAIL PROTECTED]>
>To: <[EMAIL PROTECTED]>
>Sent: Friday, July 05, 2002 6:48 AM
>Subject: Re: [JAVA3D] Trouble with Link/SharedGroup - is it a Java3D bug??
>
>
> > Subject: Re: [JAVA3D] Trouble with Link/SharedGroup - is it a Java3D bug??
> >
> >
> > > We also use shared groups to put in trees, boulders, etc, but we haven't
> > > put all that many into a scene yet. I've not noticed any specific
> > > memory issues (Java 3D is a memory hog but we all know that).
> >
> > Sigh, ok, nobody seems to believe me....
> >
> > Here is a small code snippet that demonstrates the effect (e.g. in a
> > HelloUniverse environment).
> > I turns out that the memory consumption depends on the complexity/depth of
> > the subtree that
> > is used below the SharedGroup. I create an arbitrary tree with some
> > ColorCubes and TGs and
> > either Link/SharedGroup or cloneTree() it. The bigger the subtree, the
> > better the Link-solution
> > should be. In cases where the subtree in question only consists of one or
> > two groups, it should
> > not make any difference at all anyway.
> >
> > The samples creates only 10 clones/links. While cloning keeps the memory
> > consumption well
> > below 40megs, Link gives an OutOfMemoryException even with -Xmx256mb.
> >
> > A smaller test (createDeepSubGraph(6)) works marginally with 256MB and
> > produces
> > 2.021.371 instances of javax.media.j3d.BranchGroupRetained.
> >
> > I'd really appreciate if someone could confirm my observation.
> >
> >
> > ================================================================
> >
> > /**
> > * Create a subgraph with 10 objects by either Linking or cloning an
> > object.
> > */
> > Node create10Objects( boolean clone )
> > {
> > Group grp = new Group();
> >
> > // Create a non-trivial subgraph
> > Node object = createDeepSubGraph( 7 );
> > System.out.println( numNodesCreated );
> >
> > // Setup sharedgroup
> > SharedGroup shared = null;
> > if ( ! clone )
> > {
> > shared = new SharedGroup();
> > shared.addChild( object );
> > }
> >
> > // Add the objects
> > for ( int i = 0; i < 10; i++ )
> > {
> > TransformGroup tg = new TransformGroup();
> > Transform3D t3d = new Transform3D();
> > t3d.setTranslation( new Vector3f( i * 2.0f, 0.0f, i * 2.0f ) );
> > tg.setTransform( t3d );
> >
> > // either link or clone the object
> > if ( clone )
> > tg.addChild( object.cloneTree() );
> > else
> > tg.addChild( new Link( shared ) );
> >
> > grp.addChild( tg );
> > }
> > return grp;
> > }
> >
> >
> > /**
> > * Create subgraph of a certain depth.
> > */
> > static int numNodesCreated = 0;
> > Node createDeepSubGraph( int depth )
> > {
> > BranchGroup g = new BranchGroup();
> >
> > // One new cube...
> > g.addChild( new ColorCube( 0.1 ) );
> > numNodesCreated++;
> >
> > // ...and maybe two subgraphs
> > if ( depth > 0 )
> > {
> > Transform3D t3d = new Transform3D();
> > Transform3D t3d2 = new Transform3D();
> > TransformGroup tg;
> >
> > t3d.setScale( 0.2 * (double)depth );
> > t3d2.setTranslation( new Vector3f( 0.2f, -0.2f, 0.0f ) );
> > t3d.mul( t3d2 );
> > tg = new TransformGroup();
> > tg.setTransform( t3d );
> > tg.addChild( createDeepSubGraph( depth - 1 ) );
> > g.addChild( tg );
> >
> > t3d.setScale( 0.2 * (double)depth );
> > t3d2.setTranslation( new Vector3f( -0.2f, -0.2f, 0.0f ) );
> > t3d.mul( t3d2 );
> > tg = new TransformGroup();
> > tg.setTransform( t3d );
> > tg.addChild( createDeepSubGraph( depth - 1 ) );
> > g.addChild( tg );
> > }
> > return g;
> > }
> >
> >
>===========================================================================
> > To unsubscribe, send email to [EMAIL PROTECTED] and include in the
>body
> > of the message "signoff JAVA3D-INTEREST". For general help, send email to
> > [EMAIL PROTECTED] and include in the body of the message "help".
>
>===========================================================================
>To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
>of the message "signoff JAVA3D-INTEREST". For general help, send email to
>[EMAIL PROTECTED] and include in the body of the message "help".
Lan Wu-Cavener
Dept. of Landscape Architecture
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
