Re: [JAVA3D] MathEngine is " Windows only " : Y U C K !

[Anselm Hook]

Hmmm, I'll comment on the below message since I'm working away on various projects here at MathEngine including the OMG stuff.   First of all, Java3D is not intrinsically platform independent - you have to ask the vendor for a port to your platform.  Fundamentally it's no more nor less platform independent than MathEngine.  J3D has to bind to target hardware on each platform, and MathEngine has to support numerical methods optimally on each platform.  The PS2 port of MathEngine for example is a total rewrite optimized at the assembler level for the rather unique characteristics of that device.  There exist implementations of Java3D for Windows and Linux and Solaris and Irix.  MathEngine runs on those platforms and on the PS2 - albeit not yet in commercial release.  Secondarily Java3D deployment is still not mature.  Deploying J3D based applications requires more than a single click, and this makes J3D untenable for web based content.  The problem fundamentally is an idealistic attitude at Sun, which eschews interoperation with legacy frameworks such as Windows in favour of their own entire model of reality.  Much of the good technology coming out of the Sun effort is difficult to install, difficult to deply.  Technology is irrelevant if deployment is difficult.  Thirdly, the first iteration of MathEngine technology was free to download at the very least - nobody else has put themselves up to that kind of general scrutiny including Ipion.  The proposition has always been to only charge people for genuine value - if there was genuine value and a genuine relationship then dollars should be negotiated.  Anyway, direct scrutiny has resulted in intense technical feedback and intense improvement to MathEngine as a result.  The new engine is orders of magnitude faster, energetically stable, much smaller, and uses entirely new algorithms and methods not based on Baraff.  GDC 2000 (www.gdconf.com) will be the place to see the next iteration of all this.   I think everybody agrees on and believes in the fundamental value of addictive web based multiplayer games with rich emergent behavior.  We've all have a vision of making that kind of killer web app - it's kind of a holy grail.  I actually suspect that interactive experiences / games will be the dominant form of entertainment and education within a decade and that cinema will fade away.  But ever since starting to work with dynamics last year I've realized that graphics isn't really as important as underlying behavior in achieving these kinds of goals.  A formal approach to behavior systems based on newtons laws could cover most of the typical cases that people imagine when they imagine building immersive experiences.  Best of all it would allow different game developers to make different objects that actually could interact together.  We've never had a "lingua franca" for describing the interaction of behavior systems.  Many efforts to provide game building toolkits have failed because different models of object behavior and interaction were insufficiently compatible with each other.  I believe that fundamentally most objects can interact adequately through physical properties such as position, collision and the variety of other forces and constraints that we are so used to perceiving as human beings.   Perhaps the place this is all going to happen first is the PS2.  It may very well be the first umbiqitous computing platform.    - Anselm Hook   (Note that we visited Sun and talked about the points below ( that we provide Java Dynamics for them for free).  They suggested that we go through the java community standards approval process, which we may do but haven't yet committed to.  We also wanted a port of J3D to PS2 but it turns out that this is harder than first appears so that is also on hold.  So solutions today are still somewhat fragmentary.  It's clearly our ambition to solve this and everybody is keenly aware of these issues. )   -----Original Message----- From: Discussion list for Java 3D API [mailto:[EMAIL PROTECTED]]On Behalf Of Matthew Gahan Sent: Sunday, February 06, 2000 10:50 AM To: [EMAIL PROTECTED] Subject: Re: [JAVA3D] MathEngine is " Windows only " : Y U C K ! Mathengine isn't the only runner in this race, www.ipion.com should be considered a major contender. Not least because Ipion's costs a tenth of what Mathengine's kit costs ($5000 ~ish for Ipion with some royalty's...). According to the doc's this is a highly modular native c/c++ library, which I am guessing would lend itself well to JNI integration... The Open Media Group managed this with Mathengine. The results were ok...but still slugish by current standards...   Ipion has a very low overhead (20% cpu usage estimate). It also has some very nice optimizations for dealing with very large numbers of Free body systems (ideal for particles).         -----Original Message----- From: The Casteels <[EMAIL PROTECTED]> To: [EMAIL PROTECTED] <[EMAIL PROTECTED]> Date: 06 February 2000 15:19 Subject: Re: [JAVA3D] MathEngine is " Windows only " : Y U C K !   P. Flavin wrote: Java3d is a portable rendering solution. MathEngine is a non-portable thing limited to rendering on a single OS : [.......]   Java3d is about  RENDERING   ( with collision detection ).    Pure Java is portable.  Be portable.  Be on the Web.  Be everywhere. With the exception of Java3D being about Rendering (a limited view of Java3D's potential), I agree with everything you say, and I think Java3D is the greatest,  however without dynamics (mathengine) and isosurface generation (vtk) built into the core processes, Java3D can not realize it's true potential. Although I'll admit that I hadn't considered that mathengine and vtk wouldn't be portable. Perhaps I wasn't clear in what I was suggesting.  I would like nothing more than to be able to accomplish what I need with 100% pure Java. What I was really after is development of Java packages like, JavaDynamics3D, JavaParticles3D, and JavaVolume3D. I'm suggesting that since MathEngine and VTK are open source, most of the work is already done. Imagine how cool it would be to be able to: Build a web based multiplayer pool game with realistic dynamics. Render smoke, fire, and explosions. Animate fluids, like tipping over a can of paint. (or two can's of paint and watch the colors blend) Animate cubes of Jell-O falling onto a surface, jiggling and bouncing. At this point I can't do those things without using mathengine or vtk. (or spending months figuring out how to write it in java, then ending up with something that's really slow)