> From: "Scott Nelson" <snelson@canopus.llnl.gov>
> Date: Mon, 5 Dec 1994 11:19:15 -0800
You leave for time off, and you get ... mail avalanche.
> Based on the message traffic it seems that there really
> are two groups of people:
>
> 1. a Doom like interactive, VR, linked, network/info browser
> 2. a complex scientific 3D data object visualizer
>
> The portability requirements for both groups are the same, it's
> just that #1 contains a few complex objects while #2 contains
> many simple objects.
?? I don't understand. You mean tall narrow heirarchies, or wide,
broad ones ??
> We had looked into making a combined
> group that did both but preliminary examinations were not
> encouraging.
In what way? Can you detail?
> Perhaps, it's time to break these into two groups? The
Why? I don't understand the logic for doing this.
> more that I look into my effort to make 3D distributed
> objects, the more it looks like VRML may not be the answer
> (it seems to be too generic for my needs --
In what way? too many bells and whistles that you don't
want to bother with?
> I need something
> that knows more about the data since there will be A LOT
> of data compared to typicaly VRML scenes).
What do you mean "know more about the data"?
Do you mean geometric primitives like a height mesh?
e.g.
"Here is a hieght mesh. Only z values are given. The x & y values
are implicitly integers, on a square grid" ?
You can do very powerful operation on data, even when you are not
geven any hints about is structure or connectivity. Take a look
at the IBM IX product -- "Interactive Explorer" -- it boasts
ability to visualize, at 20-30 frames a second, of data sets with
10 million polygons in them. Basically, it engages in tricks by
doing automatic mesh/connectivity simplification. Although this
product doesn't (and may never) understand vrml, the basic
algorithms and concepts are portable, and I know of nothing in vrml
that would prevent application of those types of algorithms.
> Any feedback on these thoughts would be appreciated.
--linas