Re: Re[2]: [Threedsia-dev] Question or Two

[Peter S. <ka...@gm...>]

Hi folks,
Lars wrote :
> ...
> Yes. Because it would be SIMPLER but also powerful enough. I've read
> an article from a SGI guy that explains why VRML is the wrong why to do
> things. For one, you have to get the whole scene even if you are not
> near that object. Also VRML is NO standard it has been polluted by $GI
> and M$ to be unusable. Also it is very big and it is very hard to
> convert "normal" 3D-data in VRML. It has no normed support for sprites,
> particles or a mixture of Line and Polygon representation (Ever tried
> to render a web of a spider in VRML). It has a very bad handling of
> polygons etc. It also has no support for voxels or volumetric models.
> there is no possibility to do REAL spotlights or shadows or
> bump-mapping or moving textures or...
> Also it is very noisy and not binary.

I can accept that these technical arguments are true (but i can not
confirm them).  Now look at the reality, i tested vrml a month ago with a PIII 500
and an (opengl supported) 3D card over an ISDN line.  WELL IT DROVE ME
CRAZY. First, it took a VERY long time to load the whole scene, i saw the object
apearing one for one, but couldn't do anything. then when i tried to move
it was at a rate of 1 frame every 5 seconds.  I don't know on wich PC they
tested this scene but i think it was some BIG linux cluster :).  I have seen
less complex scenes which rendered smoothly, but the matrix will be complex,
so it is no reference...
The fact is, i dont know who invented vrml and i dont know the specs, but
i do know that it cant handle high complexity. I think this has to do with
the no LOD (= Level Of Detail) support.

Anyway VRML seems not the way to satisfy the user (i mean they get
frustrated).  I've seen opengl and i like it (i also think that the future lies in
opengl, but thats IMHO)

peter


-- 
Sent through Global Message Exchange - http://www.gmx.net