To: "Martin Baker"
Subject: RE: [www-vrml] simulations, physics and game generation
Date: Thursday, September 21, 2000 6:07 PM
Martin:
Some thoughts..
=>
=> I realise I can use the PROTO mechanism for my own program, but I was
=> thinking of a standard that other people might use.
=>
As you suggest, PROTOs are our current construct for node
creation. HANIM, for example, represents a use of PROTOs
for the representation of humans. Numerous other PROTOs perform
similar duties. I don't see this as an issue of the scene
graph itself so much as a higher-level issue of building on top of
scene graphs. That is to say---extensibility. In the long term,
we should probably be looking at XML schemas as representational
vehicles.
If existing 3D game engines were to move in the direction of
supplying an XML representation or building upon X3D, it would
go a long way toward convergence and to better surfacing the
relationship between geometry, dynamics and extensibility. I
won't get into the issue of whether or not they should use DOM
because I don't know enough about the relative performance
issues, but the representational benefits of XML documents as readable
file structures could be very useful. Imagine being able today
to take a Quake3 level, along with bots and scripted behaviors,
and being able to easily browse it using an XML editor. It would
be wonderful.
-paul
=> -------------------------------------------------------------------------
=> for list subscription instructions,
=> send email to www-vrml-request@web3d.org with text "info"
=>
=>
From: "Martin Reddy"
To: "Martin Baker"
Cc: <www-vrml@web3d.org>
Subject: Re: [www-vrml] simulations, physics and game generation
Date: Friday, September 22, 2000 1:53 AM
Hi there Martin,
This sounds like interesting work. Perhaps you might want to consider
starting a working group if there is enough interest with people on the list.
This might provide a good forum for more directed discussion and developments
on this issue. There is a rather out-dated document on the Consortium web
site that describes the process for creating working groups:
http://www.web3d.org/fs_workinggroups.htm -> Process
This page should be updated soon to describe the new Team structure rather
than the old TAB process. However, it provides a good start for the time
being.
Cheers,
Martin.
---------------------------------------------------------------------------
To: "'Martin Baker'"; "Martin Reddy"
Cc: <www-vrml@web3d.org>
Subject: RE: [www-vrml] simulations, physics and game generation
Date: Friday, September 22, 2000 2:31 PM
Martin and Martin,
I too think the same way - as of now Avatars are, kind of, walking on the
space - not on earth and are all pervading. I too won't be able to join any
physical meeting because of the same reason - but would like to be in the
working group.
/Biswajit
> -------------------------------------------------------------------------
> for list subscription instructions,
> send email to www-vrml-request@web3d.org with text "info"
From: "Paul Fishwick"
To: "Martin Baker"
Subject: RE: [www-vrml] simulations, physics and game generation
Date: Friday, September 22, 2000 3:41 PM
Martin:
I think the other Martin was right -- probably you
starting a discussion group under X3D:
X3D for physics, animation, etc..
would be most welcome.
-paul
-------------------------------------
Paul Fishwick, University of Florida
CISE Department, Bldg. CSE 301
PO Box 116120
Gainesville, FL 32611
web: http://www.cise.ufl.edu/~fishwick
email: fishwick@cise.ufl.edu
phone + fax: (352) 392-1414
From: "Aaron E. Walsh"
To: "Martin Baker"
Cc: "Martin Reddy" ; <www-vrml@web3d.org>
Subject: Re: [www-vrml] simulations, physics and game generation
Date: Friday, September 22, 2000 5:42 PM
Hello everyone,
At the moment work on exactly this is going on between the Web3D
Universal Media working group and the MPEG4 Animation Framework group.
Martin, I'll forward your message (below) to Mike (Mikaël
Bourges-Sévenier) since it's on topic for the work already underway.
To
join in please subscribe to either list/reflector described at:
http://www.web3dmedia.com/UniversalMedia/conferences/siggraph2000/
Here you'll find a description of this work, as well as notes on the
joint Web3D/MPEG4 meeting held at Siggraph. Next month the outcome of
these groups will be presented to MPEG4, as the Animation Framework has
been accepted as a development project for version 5 of MPEG4 and will
be the framework upon which Web3D Universal Media dynamic filters are
built.
Regards,
Aaron
--
-------------------------------------------------------------------
Aaron E. Walsh http://www.mantiscorp.com/people/aew/ 617.350.7119
-------------------------------------------------------------------
From: "Mikael Bourges-Sevenier" <mikael@ivast.com>
To: "Aaron E. Walsh"; "Martin Baker
Cc: "Martin Reddy" ; <www-vrml@web3d.org>
Subject: RE: [www-vrml] simulations, physics and game generation
Date: Friday, September 22, 2000 7:00 PM
Dear Martin(s), Aaron,
As Aaron said, we are currently working on a proposal for MPEG-4 called "A
framework of animation tools". The framework is organised around 6 types
of
models (following Funge, Terzopoulos, and al. categories): geometry,
modeling, physics, biomechanical, behavior, and cognitive.
Of interest regarding your question is physics models. We propose to use
mass-spring-damp-ext forces to simulate dynamic models.
Feel free to contribute, the proposal is still in draft and your comments
(and updates!) are very appreciated.
All the best,
Mike
From: "Sandy Ressler"
To: "Martin Baker"
Cc: "Martin Reddy"; <www-vrml@web3d.org>
Subject: Re: [www-vrml] simulations, physics and game generation
Date: Friday, September 22, 2000 8:44 PM
While I'm not against a new working group and I think this is a great
topic...are people aware that there exists a lot of stuff in this domain that
is
commercially available. The past two months Sept, Oct of Game Developer Magazine
had a series looking at physics SDK's. So there is a lot of stuff available.
I
just think it's important to have clear goals AND I also think it's important
to
have commercial requirements ala MPEG. What's the business case? and what's
wrong with existing solutions?
Sandy
Martin Baker wrote:
> -------------------------------------------------------------------------
> for list subscription instructions,
> send email to www-vrml-request@web3d.org with text "info"
--
..................................................
Sandy Ressler
About.com Guide to Web3D
http://web3d.about.com
web3d.guide@about.com
..................................................
Sign up for my free newsletter at:
http://web3d.about.com/library/blnews.htm
..................................................
From: "Aaron E. Walsh" <aaron@mantiscorp.com>
To: "Sandy Ressler" <web3d.guide@about.com>
Cc: "Martin Baker"; "Martin Reddy"; <www-vrml@web3d.org>
Subject: Re: [www-vrml] simulations, physics and game generation
Date: Saturday, September 23, 2000 12:23 AM
Sandy Ressler wrote:
> ... So there is a lot of stuff available. I just think it's important
> to have clear goals AND I also think it's important to
> have commercial requirements ala MPEG.
Hi Sandy,
Agreed 100%. Mike and I had talked several months ago (prior to our
joint animation meeting at Siggraph this July) about setting up a Web3D
Animation working group because of the work going on in MPEG4 and the
Web3D Universal Media groups (ie., the Web3D Animation Framework +
UMedia Filters), but decided to put it on hold for a while until the
work in our own groups has taken root and merits a new wg in the Web3DC.
I imagine that we'll actually propose a Web3DC Anim Working Group later
this year, perhaps after the MPEG4 meeting next month, but our current
working groups were plenty to get the ball rolling... to boot, the
Animation Framework is taking into account several existing technologies
rather than creating them from scratch (NURBS from blaxxun X3D proposal,
for example, persistent media cache from our Universal Media URN work,
etc.). Once this gels through MPEG4 and Web3DC UMedia I think it'll be
time to create a unified Web3D Animation Working Group, but not until
then for the reasons you mention below.
Cheers,
Aaron
--
-------------------------------------------------------------------
Aaron E. Walsh http://www.mantiscorp.com/people/aew/ 617.350.7119
-------------------------------------------------------------------
> -------------------------------------------------------------------------
> for list subscription instructions,
> send email to www-vrml-request@web3d.org with text "info"
From: "Chris Marrin"
To: "Martin Baker"
Cc: "Paul Fishwick"; <www-vrml@web3d.org>
Subject: Re: [www-vrml] simulations, physics and game generation
Date: Saturday, September 23, 2000 4:58 PM
Martin Baker wrote:
>
> ...
> For instance:
> 1) I have found the need to have lots of copies of sub-sections of the
> scenegraph, but with a few differences. For example a particle system,
each
> particle needs to be exactly the same but with a few differences (position,
> velocity, possibly colour). DEF/USE won't allow differences, duplicating
the
> substructure is inefficient.
This is a good point and you are right that VRML can't really support
it. I am adding a "shared" facility to Blendo which should help. You
would do this:
PROTO MySpinningBox [
shared SFChildNode scene Shape {
appearance Appearance {
material Material {
diffuseColor 0.5 0.2 0.8
}
}
geometry Box { }
}
DEF T Transform {
children FROM scene
}
TimeSensor {
loop TRUE
fraction DO { T.rotation.angle = fraction*Math.PI*2; }
}
]
Here's what will happen. When you create the first instance of this
PROTO, you will get a "scene" field containing the Shape node. This
gets
routed to the Transform and you see a spinning box. Whe you create the
second instance, the scene field is reused and the same Shape is routed
to the new instance's Transform. As you create more boxes, each will
have its own Transform and TimeSensor (presumably there would be an
interface to change the spin rate, which I did not show).
The only way to do this in VRML is to make the Shape hierarchy global
and pass it as a field to each instance, which is not object oriented
and is fragile.
The above is a simple example, but you can imagine more complex ones
where the use of a shared field would result in one copy of a very
complex scene, perhaps with many interfaces.
> 2) Say we are simulating a rock hitting something and breaking into 2
parts.
> Then we need to dynamically remove one sub-graph and add two different
> sub-graphs, I think this needs to be made more efficient, so that we can
> define a fixed size pool of sub-structures, for example.
Using a shared field, you could create a shared "rock manager" which
would be supply rocks to new instances from a free list, and would add
rocks to the free list when an instance shuts down.
>...
> I think what I'm suggesting is to make the standard a bit more object
> oriented. ie split out the node definition from the multiple instances
that
> will be created at runtime.
Let me know if you think such a capability would solve your problems.
--
chris marrin ,""$,
cmarrin@arch.sel.sony.com b` $ ,,.
(408) 955-3049 mP b' , 1$'
Sony ,.` ,b` ,` :$$'