If you do decide to pursue approach 2) that you describe below, you
may want to take a look at the "display facing" technique covered in
"3D User Interfaces with Java 3D". The latest version of the code
that comes with the book, available free at
www/manning.com/barrilleaux, has a revamped version of a view change
detection system that does not rely on behaviors. It is intended to
drive such view dependendant actions as display facing, constant size,
and display overlay. Take a look at the DisplayFacing example and its
attendant code.
--jon
> From:
> Shawn Kendall <[EMAIL PROTECTED]>
>
> Thu 11:19 AM
>
> Subject:
> Improving our Particle System - third repost
>
>
>
> We have a very basic particle system running for Java3D.
> I have a question on how to improve it's performance but first here are few things
>we learned that may be of some use to others out there.
>
> 1) OrientedShape3D's can not always be used for Particles. When we first started
>this, we use OrientedShape3D's, but it turns out that
> because the orienting effect happens at the Shape3D level, you can not also have a Z
>rotation of the Shape3D because any and all
> rotations higher in the scene are overwritten by the OrientedShape3D. This is of
>course the point of using OrientedShape3D's, they
> always turn to face the view. The problem is that we need local Z rotation and
>turn-to-face the view rotation. For smoke effects in
> particular, you need it to get a "billowing" effect. So we tried using the
>Billboard behavior with worked fine because we could set the
> behavior higher in the scene than the Z rotation. But it does not always update in
>the same frame as the view so it looks jittery from time
> to time. This is a known problem and one of the reason OrientedShape3D was created.
>Alas, it's great for trees but not so for all particle
>
> effects.
>
> 2) We are using separate TransformGroups, Shape3D's and Appearances for each
>particle! This is the most expensive way, but for the
> test pass, we agreed it would be OK. However, now we wish to give a try at
>improving this by combining all the particles for a single
> system into one Shape3D and Appearance. Color/alpha, coords, and texture-wise this
>works out fine because TriStripArray give us
> enough variation to get a good particle effect, i.e. changing position, color,
>alpha, but same texture. The problem is the orientation of the
> primitive. It needs to always face the camera, just like a Billboard behavior or
>OrientedShape3D. Now we are thinking of writing our
> own view transformer that would xform each triangle when it does the particles
>update. This isn't too bad, still not that cheap, but I can't
> even think of another way. So....
>
> The question is, what is the best/fastest way to xform all those triangles to the
>view? Is there some "cheat" we can do to just tell it to
> draw them right to the screen without doing any world to view transform, bypassing
>the need to xform them back to the view. Or maybe
> the Raster Geometry is an option? We considered doing a few tests with that but we
>except it doesn't really do what we want. Can
> Rasters scale and rotate?
> Thanks for any help.
>
--
__________________ JMB and Associates, Inc. ___________________
Jon Barrilleaux 3800 Lake Shore Ave. Consulting for 3D
[EMAIL PROTECTED] Oakland, CA 94610 Web Applications
510.444.0231 vox & Tech Industry
510.444.0231 fax www.jmbaai.com Management
===========================================================================
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".
