Florin,

It was my understanding that geometry by reference simply allows one to
hand geometry to a shape3d, without requiring the shape3d to make its own
internal copy.  Therefore, it eliminates the need for a copy operation.

I dont believe that any indirection is added.  A reference to geometry is
always needed somewhere, regardless of if that geometry is a copy or an
original.

Therefore, I dont believe that by reference, performance is negatively
impacted.

Further, I dont believe this has anything to do with updating geometry.
Geometry can always be updated, unless the branch group is compiled.

Thoughts?

Ian

On Mon, 6 Jan 2003, Florin Herinean wrote:

> Using geometry by reference only makes sense when you want to modify the
> geometry at runtime, otherwise you just introduce another level of
> indirection and hinder possible optimizations.
>
> Cheers,
>
> Florin
>
> -----Ursprüngliche Nachricht-----
> Von: Ian M Nieves [mailto:[EMAIL PROTECTED]]
> Gesendet: Montag, 6. Januar 2003 09:55
> An: [EMAIL PROTECTED]
> Betreff: Re: [JAVA3D] size of transform3D and transformGroup
>
>
> Hans,
>
> Thanks for your reply.  Your point is very clearly made.
>
> I may be in trouble though.  I might want to be able to share geometry,
> but have different appearances.  I will have to look into this more.
>
> Another question though.  Is is possible for me to use geometry and
> normals, by reference (instead of by copy), and have each reference point
> to a single, shared static version of the geometry and normals?  Would
> that have any effect?
>
> Thanks,Ian
>
>
> On Sun, 5 Jan 2003, Hans Horn wrote:
>
> > Ian,
> >
> > I don't know about the individual memory reqs of Transform3D and
> > TransformGroup, but I do know from my own projects by experimenting with
> > SharedGroups, that the SharedGroup approach is the way to go with clear
> > advantanges in the construction time of the scene, the memory footprint
> and
> > the performance of the final scene.
> >
> > In one of my previous projects I built a J3D molecular viewer where I can
> > turn on the use of SharedGroups for
> > atoms (Primitive=Sphere) and bonds (Primitive=Cylinder) explicitly.
> > When loading a 1500 atom molecule, the whole app requires about 30MByte of
> > memory with SharedGroups, but without I quickly run out of the 128MByte
> > memory I give to the JVM.
> > For a smaller molecule of say 100 atoms, where I can get thru with doing
> it
> > both ways, I find the construction time with SharedGroups about half
> > compared to when using no SharedGroups. Also, the scene is much more
> > responsive to mouse operations when using SharedGroups.
> >
> > I found, however, that SharedGroups have their limitations, too. If you
> need
> > to be able to pick an individual object in the scene and do some special
> > rendering (e.g. change its appearance) on it depending on its picked
> state,
> > it must not be in a SharedGroup, as all items in the group share the same
> > appearance.
> >
> > H.
> >
> >
> ===========================================================================
> > 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".
>
> =====================================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".

Reply via email to