Hi Dirk,

> > what about Shadow Volumes in OpenSG 2.0? If I
> > remember correctly someone was suggesting that
> > as a new feature for OpenSG 2.0 in the past.
> > Is anybody actually planning to add that? I saw
> > that feature in GizomSDK, very nice!


> In general shadow maps are more efficient (IMHO), unless you put serious
> effort into optimizing the shadow volumes. There was  anice paper at the

Yes they are indeed very efficient. But we often run into a 
quality problem. The worst things are huge scenes containing
very detailed objects. Consider a supermarket, you'll end up
mapping a shadow map pixel on a whole product within the 
supermarket.... I whish I had shadow volumes in such a case
(and I won't care for frame rates).

> I3D a few years ago that looked very good, but quite a bit of effort.
> The only big advantage is the quality of volumes, they are very sharp
> shadows. But with good map algorithms like the new ShadowViewport you
> can get pretty good quality, too.

Yea, of course that's not easy to do, I know... Was just checking if
there's any serious effort/project in implementing that.

> > http://www.gizmosdk.com/
> >
> > Speaking of gizmo... Did you see the new "clustering"
> > support? It's not as smart as OpenSG's but they are
> > catching up!
> 
> I couldn't find anything about that. They do support multipage TerraPage
> servers now, but that's just for data paging and has nothing to do with
> the scenegraph per se. Can you point at some information about that?

Sorry, it's apparently still not official. You can see it in
the news section for the beta version:

15 March 2006
GizmoSDK 1.3 beta 20:
* Distributed Scene Graph
* Extensive TerraPage updates
* Multi Host databases
* OFL Mesh format updates
* New font system


I exchanged some emails with Anders Modén, the CTO of GizmoSDK. BTW
he gave some interesting interviews where he mentions OpenSG and OSG,
read more here:

http://www.modsim.org/Articles/GizmoSDK/GizmoSDKInterview.pdf#search=%22Ande
rs%20Mod%C3%A9n%20opensg%22

Anyway, in those emails he roughly explained to me the distributed scene 
graph concept, here are two snippets:

---------------------------
> > That sounds like a lot of work, I'll probably have to look at an 
> > example. Do you have some very small tutorial example about this 
> > topic?
> 
> Nope. But here is how you do it..
> 
> 1. Create and instance of a gzWindowSyncControl on the main machine 2. 
> issue gzWindowSyncControl::run() to start a demo loop of 
> (startRender() and swapBuffers())
> 
> ..
> 
> 3. Set up a gzWindowSync instance on each render hw 4. issue 
> gzWindowSync::addWindow(your window); 5. issue gzWindowSync::run()
> 
> ..
> 
> Now you can swap all clients using the 
> gzWindowSyncControl::(startRender()
> and swapBuffers()

---------------------------------------------------------

> Sounds good. So the geometry is distributed transparently?

Not by default. You have a couple of options to do it.

1. Set up a web server or a purl server (a gizmo runtime data 
distributor) and use http or purl urls in your filename of the scenegraph

2. use GizmoDistribution to "pack" graphics data into distributable objects.

For a non dynamic world the alt 1 is the easiest way. But then you will not
be able to control rotations, animations etc for distributed objects.

By using option 2 you can set the rotation of a node in the scene graph and
it will be transferred into a distibuted call that will update the scene on
all displays. This is the true distributed scene.

------------------------------------------------------------



-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to