[vos-d] parameterized views

[Karsten Otto]

Hi folks!

 From the top of my head, I'd say this idea of yours sounds cool, but  
is probably going too far. After all, the idea is to present a  
synchronized world for everybody to *share*, i.e. Gibson's  
"consensual hallucination".  If every user gets to see the world in  
an entirely different way, you will ultimately wind up in a situation  
where everybody can only work in its own private (view of the) world,  
and nobody understands anybody else anymore...

A: "Wow, that vobject over there looks so cool."
B: "What vobject? I dont have that in my AOI list."
A: "Its right there... what stupid kind of AOI management do you use?"
B: "*My* AOI is fine, thank you very much. And who are you anyway, I  
don't see you!"
A: "Huh? I'm staring right in your face."
B: "I don't have to listen to rude voices in my head. Go away."

Despite this extreme example, everybody *will* see the world in his  
or her or its own way anyhow, since every client may have a different  
kind of  access to the VOS world. One may use libvos, someone else  
JaVOS, someone else a python client... And while a human user will  
want to see pretty animations, an agent likely does not really care  
about individual keyframes. So, everybody will get a form of your (b)  
already if they want to.

Regarding (a), I don't see how a generic server could possibly  
support all weird kinds of queries a client may want to formulate.  
For example, assume your are at x=0. If a parent object P is at  
x=-20, but a child object C is at *relative* x=+15, do you report C  
in the 10 units AOI query? It is only 5 units away from you... but  
then, what about P? If you leave that out, C does not make sense! You  
may easily wind up in a situation like the conversation above.

Bottom line: Let's not encourage divergent behaviour to a point where  
it gets schizophrenic, rather try to support a strong consensus.

Regards,
Karsten (kao)