Re: [Opensg-users] Problems loading VRMLs models

[antonio_lioy <ant...@li...>]

 
> I'm not sure I follow entirely. Are you trying to load several parts 
> (of, say a car), then want them connected in your app under the root?
> If so, you naturally need to connect them yourself, and try to recover =

> the connectivity somewhow. How to do that varies wildly depending on 
> your data, if you just have a car (i.e. body + wheels, not suspension) =

> it's should only be a matter of adding all parts to a transform and mov=
e 
> that around.
> 
> If the transform is identity, then you might have the data in the 
> vertices (that are then relative to world center, not object center). 
> Check your data and make sure you export stuff correctly and get what 
> you expect in the VRML.
> 
> (I often find this to be the problem, that the VRML-data is not exactly=
 
> what I want. OpenSG does a very good job at importing the VRML keeping =

> the exact data and structure. I haven't yet found a flaw in OpenSG w.r.=
t 
> this, it is usually my data.)
> 
> Regards
> /Marcus

Thanks for your answer, but my problem actually is that I have a complex =
model drawn with a CAD application that is split into many components by =
the CAD app. Each component is saved to a single vrml file. If I just loa=
d all the models under a common parent in the scenegraph I can correctly =
see the big model cause in each VRML file the CAD app saves a matrix that=
 specify where each model should be with respect to the world axis. In my=
 application (which is a stereo viewer in a CAVE environment)I need to pi=
ck a single part of entire model through a tracked pointer. When a pointe=
r picks a model, the transformation performed by the pointer (which I com=
pute through tracking system) should be reproduced by the model. My probl=
em is now that when I move the objects with the pointer they move keeping=
 their  initial transformation, that is objects centered in origin are mo=
ved corretly, and objects transformed keeps an offset from the point that=
 I actually move. For now I set to identity the local transform when I pi=
ck an object but for some models I cannot find  this transformation where=
 I expect it to be, so they continue to mantain their offset.
I hope that all this now looks clearer...
Thanks

Esimio prof. Antonio Lioy