openvrml-develop — OpenVRML developers mailing list

[openvrml-develop] request for comments : a perl module for openvrml

[clayton c. <dr...@sm...>]

heya:

with the new messages
about progress updates ive 
had a hint of inspiration
and i want to try something new with perl

if all goes well it should be beneficial for all


i noticed matt s of axkit fame made those nifty perl modules
for wrapping around the libxml c libs XML::LibXML

when braden mentioned:
   Once these changes are in place, fully generic traversals of the
scene
   graph will be possible.

i got excited and thought maybe i should try my hand at hooking up
some perl modules to the c libs of openvrml

mostly for using to parse and validate vrml

a heady task im sure
&
im really not sure what i need to know

def some header file & libs info
will be needed 

right now i mostly wanted to give a heads up 
and see if anyone else has ideas 
on other possible libs to 
try this with for what every reasons

anyways anyone interested let me know

im just starting to learn how  perl XS  works
so dont expect a release tomorrow!


ttfn

[openvrml-develop] Progress update

[Braden M. <br...@en...>]

Henri Manson and Tom Flynn have been quite busy getting Script node Java
working, so we can expect to see some good progress there in the next
release. It's been a while since I checked much in of significance, so I
thought folks might want to know what's up...

The main things I wanted to accomplish for the next release were to
apply a sane ownership scheme to VrmlNamespaces, and to revamp the
setField/getField stuff to be a bit saner. These are elements that stand
to be substantially impacted by the rearchitecture, so I am pressing
forward with that.

Here is the meat of the new architecture: each node implementation must
support three interfaces: NodeClass, NodeType, and Node.

        * There is one NodeClass object per scene per node
          implementation. It is a factory for NodeType objects, and it
          is a convenient place for scene-specific information that
          should be shared by all nodes sharing the same implementation.
          (Think of bound node stacks.)
        * The NodeType object is a factory for Nodes, and it is
          responsible for making available type information for a node:
          the node type ID, and what eventIns, eventOuts, fields, and
          exposedFields it has. There is one NodeType object per scope
          per node type.
        * Finally, the Node object corresponds to a node instance.

Most of the initial required work involved in moving the existing node
implementations to this architecture is done at this point. But part of
the fallout from these changes and my changes to VrmlNamespace is that
the implementation of PROTOs needs basically to be redone. I am working
on that now.

When I'm done with all this, the semantic behind getField() will have
changed a little. Right now, getField only works for exposedFields and
eventOuts. The new implementation of getField() will work for
exposedFields and fields. This is consistent with setField().
getEventOut() will provide a general mechanism for getting event values,
whether generated from eventOuts or exposedFields.

Once these changes are in place, fully generic traversals of the scene
graph will be possible.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Some Questions

[S.K.Bose <bo...@pa...>]

On Mon, 24 Sep 2001, luc...@li... wrote:

> Last i have some specific questions for U:
> 
> - How to get the global modelview transform that affects a geometry? Is
> there a quick way or should i build an appropriate NodeVisitor?

You will get accumulated modelview transformation matrix for each nodes in
RenderContext. This is calculated during render traversal.
        
> Can the getParentTransform help?

This will not help you anyway. This is applicable only for Viewpoint node
and some sensor nodes. Though there are problems with this as a node can
have multiple parents (DEF/USE). 

Thanks,

Bose

[openvrml-develop] Some Questions

[<luc...@li...>]

After working on my paper stuff and having a speech in which i 
presented my work during the year i am now ready to work again with 
openvrml. 

First of all let me express my regret for what happened the 11th of 
September. Let me say that all Italy is solid with USA and that our 
government is ready to join in the battle against terrorism.

Going back to openvrml i had the permission to release as opensource 
the project i am developing and in which openvrml plays a very big role.
Ill create a sourceforge site as soon as i get something really 
working, perhaps in one month.

Last i have some specific questions for U:

- How to get the global modelview transform that affects a geometry? Is 
there a quick way or should i build an appropriate NodeVisitor? Can the 
getParentTransform help?

- Including the following headers:

#include "Vrml97Parser.hpp"
#include "VrmlNamespace.h"
#include "VrmlNodeType.h"
#include "VrmlNodeProto.h"

gives me problems in an MFC application, while it doesnt with a console 
application. Compiling the MFC application i get errors in the 
   vc98\include\memory header, in the code defining new. 
Anyone has a clue?


Thanks in Advance
Luca

[openvrml-develop] How know the trasformations of an object.

[Daniele <dan...@ti...>]

I have written my VrmlNodeVisitor.
Now, I have to read a Vrml file and to build from this a representation =
of the world (I have to use a particular data structure). I want to know =
how take the objects from the world.

My actual problem is to know the transformations to apply to every =
single object (primitive).

The data structure I have to use comprends Class for the objects, like =
Box, Sphere, Cylinder, Cone..., moreover I can built aggregate of =
objects.

Thanks in advance
Daniele Marchionni

[openvrml-develop] ANN: FTP files moved to Web server

[Braden M. <br...@en...>]

SourceForge is phasing out FTP service. The files we had available via
anonymous FTP are now available at <http://openvrml.org/dist/>. These
files include ANTLR distributions, supplementary files for the Mac and
Windows distributions, and releases that were made before we moved to
SourceForge.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Navigation in "Examine Mode"

[Chris M. <ch...@ma...>]

Greg Roelofs wrote:
> 
> > All Cosmo did was pick a point in the middle of the bbox and use that as
> > the examine center. Then it was maintained through pan and zoom.
> 
> Ah, but the bounding box of what?  Everything visible within the view
> frustum, only those things completely contained within it, just the
> things within a half-angle view, ... ?  For example, if you have two
> balls, one centered and one to the side, you probably want to rotate
> about the centered one, not about the centroid of the two.  But maybe
> that gets too complicated with special cases.

These are all good questions. I am sure we never did anything "smart".
When we computed the bbox for use as a rotation center, we did the bbox
of everything. But we may only have done that if there was no viewpoint
in the scene. I don't quite remember...

>...It was very easy to grab the object examine all around it and
> > then return it to its original orientation. This is something I have
> > never seen in any of the other browsers.
> 
> Yes, yes, I *loved* that level of control--very nice, and most worthy
> of copying.  The only thing missing (IMHO) was a Cosmo/Live3D body-
> centered spin mode, which was very nice for manipulating building models
> (where you generally want "up" to be "up," regardless).  I can't remember
> what the other mouse buttons and shift keys did in Cosmo, but there must
> be some available combo to which one could bind the secondary mode.

The only thing we had was "straighten" which would turn the camera to
make Y up. I have seen the type of thing you are talking about. In fact,
we did a Blendo demo of a multiresolution earth globe where we used
something like that. It would always rotate about the pole and when you
went north and south, it would simply angle the camera down or up to
keep the earth in the center of the view. Of course, you could never go
further than the pole in this mode. But for viewing this type of model,
it worked pretty well.

--
Chris Marrin          | Senior Software Architect | Sony Corporation
phone: (408) 955-3049 | cm...@ar... |
[ "As a general rule, don't solve puzzles that open portals to Hell." ]
[          - excerpt from "A Horror Movie Character's Survival Guide" ]

Re: [openvrml-develop] Navigation in "Examine Mode"

[Greg R. <ne...@po...>]

> All Cosmo did was pick a point in the middle of the bbox and use that as
> the examine center. Then it was maintained through pan and zoom.

Ah, but the bounding box of what?  Everything visible within the view
frustum, only those things completely contained within it, just the
things within a half-angle view, ... ?  For example, if you have two
balls, one centered and one to the side, you probably want to rotate
about the centered one, not about the centroid of the two.  But maybe
that gets too complicated with special cases.

> The really cool thing about Cosmo's examine mode (which I have not seen
> duplicated anywhere else), is the manipulation interface. It used a
> manipulator from Inventor called a "SphereSheetProjector". This was a
> very nice class which would take the pick point and project it onto
> either a sphere or plane (sheet) depending on its distance from the
> center of rotation. Near the center of rotation, the object is
> manipulated as if your hand was on a sphere, very natural. But as you
> moved off the center, it became more and more as though you were
> dragging a piece of string along a flat surface (the sheet). The other
> end of this string was wrapped around the object, which spun along an
> axis perpendicular to the direction of drag as you moved away. It solves
> many of the problems of manipulation and makes for very repeatable
> results. It was very easy to grab the object examine all around it and
> then return it to its original orientation. This is something I have
> never seen in any of the other browsers.

Yes, yes, I *loved* that level of control--very nice, and most worthy
of copying.  The only thing missing (IMHO) was a Cosmo/Live3D body-
centered spin mode, which was very nice for manipulating building models
(where you generally want "up" to be "up," regardless).  I can't remember
what the other mouse buttons and shift keys did in Cosmo, but there must
be some available combo to which one could bind the secondary mode.

Greg

Re: [openvrml-develop] viewer->resize problem

[S.K.Bose <bo...@pa...>]

On 6 Sep 2001, [ISO-8859-1] St=E9phane Genaud wrote:

> One is : i can hardly set the image loaded by openvrml in a
>             frame with given height and width. I know the window
>             size of the client and i would like to draw the vrml
>             scene in the given limits, resizing if necessary so
>             that the full scene fits the client display.
>             The following code does not shrink (resize) the scene but
>             truncates the scene if it is too wide.
>=20
> viewer is an ViewerOpenGL object.
>=20
> extern "C"
> bool vrml_load_scene(char *szName,int width,int height)
> {
>     if(!viewer)
>         return ( false );
>     if (viewer->scene()->load(szName)) {
>        fprintf(stderr,"debug : resize to %d %d\n\n\n",width,height);
>        viewer->resize(width,height);
>        return (true );
>     }
>     else {
>        fprintf(stderr, "[server] error in VRML file parsing ...");
>        return( false );
>     }
> }

viewer->resize() will set "Viewport" only. You have to redraw the scene
after this by viewer->redraw().

Thanks,

Bose

[openvrml-develop] viewer->resize problem

[G. <ge...@ic...>]

Hi,

i am new to the list, so forgive me if it is not the
right place to ask.
I am using openvrml for a linux client-server vrml viewer :
the server loads a vrml file, and renders it in an offset
OpenGL pixmap. Several clients can then ask the server to send
them a bitmap of the OpenGL image. They also can ask the server
to rotate, shrink, etc, the image. Thus, clients without 3D cards
can see vrml files, and furthermore, with several clients, the=20
visualization is collaborative.

Currently, all this is working but i must fix some small bugs.

One is : i can hardly set the image loaded by openvrml in a
            frame with given height and width. I know the window
            size of the client and i would like to draw the vrml
            scene in the given limits, resizing if necessary so
            that the full scene fits the client display.
            The following code does not shrink (resize) the scene but
            truncates the scene if it is too wide.

viewer is an ViewerOpenGL object.

extern "C"
bool vrml_load_scene(char *szName,int width,int height)
{
    if(!viewer)
        return ( false );
    if (viewer->scene()->load(szName)) {
       fprintf(stderr,"debug : resize to %d %d\n\n\n",width,height);
       viewer->resize(width,height);
       return (true );
    }
    else {
       fprintf(stderr, "[server] error in VRML file parsing ...");
       return( false );
    }
}

If anyone could give me a hint of how to resize, that would be nice.
Thanks.

--=20
  St=E9phane Genaud                                   =20
   ICPS-LSIIT, Universit=E9 Louis Pasteur                  =20
   P=F4le API, Bd. S. Brant, F-67400 Illkirch       =20
   tel : (33)(0)390244542, 0619058113 (SFR)     =20

Re: [openvrml-develop] Navigation in "Examine Mode"

[Braden M. <br...@en...>]

On 03 Sep 2001 18:52:50 +0530, S.K.Bose wrote:
> 
> Today a bunch of fixes I have checked in the CVS. They are
> 
>         * Fixed Navigation problems in "Examine Mode", as it was
>           mentioned in
>           <http://groups.yahoo.com/group/libvrml97/message/1451>
> 
>         * Fixed some issues for bug #447971 (Navigation problems)
> 
>         * Fix for bug #213579 (Backgrounds appear to move)
> 
>         * Fix for bug #213970 (Headlight doesn't follow Viewpoint)
> 
>         * Fix for shininess component in Material node.
> 
>         * Fix for getOrientation(). In some cases it was failing.

Sweet!

I'll see if I can track down what I did to screw up the sliders in
rotation_toy. (Thanks for catching that!)

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Navigation in "Examine Mode"

[S.K.Bose <bo...@pa...>]

On Wed, 5 Sep 2001, Chris Marrin wrote:

> The really cool thing about Cosmo's examine mode (which I have not seen
> duplicated anywhere else), is the manipulation interface. It used a
> manipulator from Inventor called a "SphereSheetProjector". This was a
> very nice class which would take the pick point and project it onto
> either a sphere or plane (sheet) depending on its distance from the
> center of rotation. Near the center of rotation, the object is
> manipulated as if your hand was on a sphere, very natural. But as you
> moved off the center, it became more and more as though you were
> dragging a piece of string along a flat surface (the sheet). The other
> end of this string was wrapped around the object, which spun along an
> axis perpendicular to the direction of drag as you moved away. It solves
> many of the problems of manipulation and makes for very repeatable
> results. It was very easy to grab the object examine all around it and
> then return it to its original orientation. This is something I have
> never seen in any of the other browsers.
>

If I am right, You are talking about "Virtual Trackball". Now this Virtual
Trackball is being used for our Navigation in Examine Mode. This Virtual
trackball is Copyrighted by Silicon Graphics. Chris Morley is the best
person to tell about this hacked portion. This code was implemented by
Gravin Bell, lots of ideas from Thant Tessman and  the August '88 issue of
Siggraph's "Computer Graphics," pp. 121-129.

Thanks,

Bose
 

Re: [openvrml-develop] Navigation in "Examine Mode"

[Chris M. <ch...@ma...>]

Greg Roelofs wrote:
>...
> Also, I think someone asked about "shooting" (picking) a day or two
> ago.  That's going to be needed for a sane Examine mode, too, since
> you generally want rotation to occur about a point somewhere in the
> object or part of the object that's in front of you, preferably close
> to the center of the field of view.  I remember Chris or someone
> discussing the finer points of Cosmo's approach a few years ago...

All Cosmo did was pick a point in the middle of the bbox and use that as
the examine center. Then it was maintained through pan and zoom.

The really cool thing about Cosmo's examine mode (which I have not seen
duplicated anywhere else), is the manipulation interface. It used a
manipulator from Inventor called a "SphereSheetProjector". This was a
very nice class which would take the pick point and project it onto
either a sphere or plane (sheet) depending on its distance from the
center of rotation. Near the center of rotation, the object is
manipulated as if your hand was on a sphere, very natural. But as you
moved off the center, it became more and more as though you were
dragging a piece of string along a flat surface (the sheet). The other
end of this string was wrapped around the object, which spun along an
axis perpendicular to the direction of drag as you moved away. It solves
many of the problems of manipulation and makes for very repeatable
results. It was very easy to grab the object examine all around it and
then return it to its original orientation. This is something I have
never seen in any of the other browsers.

Since Inventor is now Open Source, perhaps someone can grab the
SphereSheetProjector class and its friends. I believe it is called
SbSphereSheetProjector, which means it is in the Basic subsystem of
Inventor. You'll need several other classes from Basic as well. I ended
up taking this class and trimming it down to just our needs. It allowed
me to get rid of some of the bloat, but you can probably take them all
to make it simpler. From what I remember, using it is pretty simple. You
set the center of rotation, and the initial pick point. Then you give it
an updated pick point and it returns a SbRotation. This is a quaternion
based rotation value that can be easily converted into a SbMatrix, which
is a simple 4x4 matrix contains just a rotation.

I can help find it if needed...

-- 
Chris Marrin          | Senior Software Architect | Sony Corporation
phone: (408) 955-3049 | cm...@ar... | 
| What is this talk of software 'releases'? Klingons do not 'release' |
| software; our software ESCAPES, leaving a bloody trail of designers |
| and quality assurance people in its wake! (unknown author)          |

Re: [openvrml-develop] Navigation in "Examine Mode"

[Greg R. <ne...@po...>]

"S.K.Bose" <bo...@pa...> wrote:

> Today a bunch of fixes I have checked in the CVS. They are

>         * Fixed Navigation problems in "Examine Mode", as it was
>           mentioned in
>           <http://groups.yahoo.com/group/libvrml97/message/1451>

>         * Fixed some issues for bug #447971 (Navigation problems)

>         * Fix for bug #213579 (Backgrounds appear to move)

>         * Fix for bug #213970 (Headlight doesn't follow Viewpoint)

>         * Fix for shininess component in Material node.

>         * Fix for getOrientation(). In some cases it was failing.

Thanks, Bose!  I believe 447971 was my jumbo bug.  I'm overrun with
contractors and remodelling at the moment, but I'll try to test things
as soon as I can.

Also, I think someone asked about "shooting" (picking) a day or two
ago.  That's going to be needed for a sane Examine mode, too, since
you generally want rotation to occur about a point somewhere in the
object or part of the object that's in front of you, preferably close
to the center of the field of view.  I remember Chris or someone
discussing the finer points of Cosmo's approach a few years ago...

Greg

[openvrml-develop] Navigation in "Examine Mode"

[S.K.Bose <bo...@pa...>]

Today a bunch of fixes I have checked in the CVS. They are

        * Fixed Navigation problems in "Examine Mode", as it was
          mentioned in
          <http://groups.yahoo.com/group/libvrml97/message/1451>

        * Fixed some issues for bug #447971 (Navigation problems)

        * Fix for bug #213579 (Backgrounds appear to move)

        * Fix for bug #213970 (Headlight doesn't follow Viewpoint)

        * Fix for shininess component in Material node.

        * Fix for getOrientation(). In some cases it was failing.

Bose

Re: [openvrml-develop] Arbitrary GLUT calls in MacLookat

[Michael L. <Mic...@hr...>]

Yuck. Sounds like a bug in Apple's GLUT port. You could delve into the 
GLUT code to see how it handles "normal" window resize to see if it does 
anything in addition to what is done when glutReshapeWindow is called? 
Sounds likes it's not passing information on to the MacOS windowing system.
  Are you using plain GLUT or GLUT with my hacks? You could try hacking 
lookat with vanilla GLUT to see if it exhibits the same problems?

Michael.

On Friday, August 31, 2001, at 06:52 , John Richardson wrote:

>
> I noticed on the Mac that if I just call glutReshapeWindow out of the blue
> (the user issues a resize window mouse click) the window resizes but then
> the Macintosh windowing system is corrupted upon exiting the program. So,
> for example, other applications can not be given focus and the login 
> screen
> gives errors. The solution is to reboot and then comment out the call to
> glutReshapeWindow.

--
Web3D, VRML, and virtual worlds on the Mac at http://www.MacWeb3D.org/

Re: [openvrml-develop] Rotating the world versus rotating avatar view

[S.K.Bose <bo...@pa...>]

On Fri, 31 Aug 2001, John Richardson wrote:

> I would like to rotate the entire scene about the x (pitch/ look up/down) ,
> y (yaw look left/right) or z (tilt / bank) axis without changing the
> avatars viewpoint. It would be relative to the VRML model that is suspended
> in the Virtual universe being viewed. Presumably this is relative to the
> local origin of the VRML model.
> 
> Any hints as to the soubroutines or OpenVRML variables to examine to
> accomplish this goal?
> 
Probably you can do it by manipulating OpenGL ModelView matrix within
ViewerOpenGL::beginObject() before rendering all nodes.

Bose

[openvrml-develop] Rotating the world versus rotating avatar view

[John R. <ric...@sp...>]

Hello,

Two Topics:

First, any hints on rotating the root node and all the children along the
x, y or z axis relative to their environment (local origin of the root
node).

A code line such as "lookat->getViewer->handleKey(x)" will rotate, tilt,
pitch the avatars view or step the avatar forward , back, slide,.... There
is also a facility to reset the viewpoint,...

I would like to rotate the entire scene about the x (pitch/ look up/down) ,
y (yaw look left/right) or z (tilt / bank) axis without changing the
avatars viewpoint. It would be relative to the VRML model that is suspended
in the Virtual universe being viewed. Presumably this is relative to the
local origin of the VRML model.

Any hints as to the soubroutines or OpenVRML variables to examine to
accomplish this goal?

John F. Richardson

[openvrml-develop] Arbitrary GLUT calls in MacLookat

[John R. <ric...@sp...>]

Hello,

I noticed on the Mac that if I just call glutReshapeWindow out of the blue
(the user issues a resize window mouse click) the window resizes but then
the Macintosh windowing system is corrupted upon exiting the program. So,
for example, other applications can not be given focus and the login screen
gives errors. The solution is to reboot and then comment out the call to
glutReshapeWindow.

When I just drag the lower left corner during normal use of the viewer
there is a resize of the viewer. I assume it is from the glutReshapeFunc
registering the resize callback which sets the gl viewport. This is
obviously fine and does not kill the mac.

OpenVRML 0.10.0. Mac OS 9.0.4.

Any comments?

John F. Richardson

[openvrml-develop] Shoot functionality

[John R. <ric...@sp...>]

Hello,

In OpenVRML 0.10.0 what are the library or GLUT subroutines and variables
to examine if I want to enable a "shoot" behavior in the viewer?

I assume the behavior would sort of conform to the "examine" mode in
Cortona and the other browsers. In addition, I would like to actually
figure out the node "picked" by the user. Thus I would be able to "examine"
the VRML model close up or just obtain information to perform some user
defined behavior.

Is the "shoot" behavior already in ViewerOpenGL.cpp?

John F. Richardson

[openvrml-develop] construct field from string

[Peter M. <me...@iw...>]

hello,

is there a nice way to construct a field from a string. I have a two
strings:
field (e.g. "center") and value (e.g. "0,0,0"). Can I update a vrmlnode
with that information
without switch (field) ...... ?

Many Thanks, peter

Re: [openvrml-develop] linux vrml plug-in for netscape

[<mk...@da...>]

>talks about this company codeweavers that allows and windows plugin to
>run on linux

Apparently a wrapper based on WINE - costs $20 , and is specifically said to
work with Quicktime and Word viewers. No mention on vrml viewers on their
pages (but they express the expectation that many plugins will work). If
anyone on this list tries it with a vrml viewer, it would be nice if s/he 
posted a quick review.

The only alternative i know is using xswallow to 'catch' and embed the X 
window of a standalone viewer in netscape.

Martin

-- 
Dr. Martin Kroeker, daVeg GmbH Darmstadt  CAD/CAM/CAQ  mk...@da...
                      Precision Powered by Penguins

Re: [openvrml-develop] linux vrml plug-in for netscape

[clayton c. <dr...@sm...>]

Christophe CERTAIN wrote:
> 
> does anybody knows a linux vrml plug-in (not a stand-alone viewer) for netscape ?
> 
> _______________________________________________
> openvrml-develop mailing list
> ope...@li...
> http://lists.sourceforge.net/lists/listinfo/openvrml-develop


http://slashdot.org/article.pl?sid=01/08/28/1552247&mode=thread

talks about this company codeweavers that allows and windows plugin to
run on linux

[openvrml-develop] linux vrml plug-in for netscape

[Christophe C. <chr...@wa...>]

does anybody knows a linux vrml plug-in (not a stand-alone viewer) for net=
scape =3F

[openvrml-develop] Win32 OpenVRML-0.11.1 and win32lookat-0.11.0 binary

[S.K.Bose <bo...@pa...>]

Compiled versions of Openvrml-0.11.1 and win32lookat-0.11.0 are available
at <http://sf.net/projects/openvrml>

Compiled versions of required libraries are available at
<ftp://openvrml.sf.net/pub/openvrml/win32_required_libs.zip>

Thanks

Bose