[Geowanking]Interdukshon

[Viveka]

At 9:55 PM -0400 3/6/03, Ben Discoe wrote:
>Hi Viveka,

Hi Ben, thanks for the response - most helpful.

>  > From: Viveka [listmail at karmanaut.com]
>>
>>  As for tech, we're using GeoVRML, X3D
>
>..ur, the same thing.  VRML/X3d is a file format.

Well, X3D is VRML200x, but not quite VRML97. The difference isn't 
just the XML and binary encodings, or the inclusion of the VRML97 
Amendment 1 stuff like NURBS and GeoVRML - there's also profiles, 
extensions, and a few spec changes that make it 
not-quite-back-compatible. To the extent that after much debate it's 
been decided (this week) that the VRML Classic Encoding of X3D will 
not use .wrl or model/vrml or x-world/x-vrml.

I suppose I should have said "The GeoVRML Profile for X3D", to be accurate.
http://www.hypermultimedia.com/Xj3D/geo.htm

>A dead one, no less.  But
>that's a tired old subject.

No flame war needed - I agree.

For me, VRML got mortally ill when Cosmo died just before SIGGRAPH 98.
http://www.frerichs.net/vrml2/requiem/requiem.html
Microsoft gassed the medics with vapourware (Chrome), and the funeral 
cortege left the building with Justin Couch in 1999.
http://www.vlc.com.au/~justin/opinion/goodbye_vrml.html

I continued to use it, even though it was dead, and even though I 
primarily use a Mac, for which there are few decent VRML tools 
available. I think I'm very stubborn, or just slow. Maybe both.

The fact that Justin is back coding away on XJ3D, the Consortium's 
open source Java X3D browser, makes the long-dead VRML even more 
useful to me :)
http://www.web3d.org/TaskGroups/source/xj3d.html

>What i wanted to comment on was:
>
>  > Right now, I'm looking for a normative list of geospatial Features.
>
>Anything on the surface of the earth, from a pebble to a continent, can be
>considered a feature.  A 'normative list' would be a major fraction of an
>english dictionary.  Perhaps you have a more constrained meaning of
>feature - in which case i could probably recommend some starting points.

Yep, all the nouns, I guess. And an infinite number of permutations 
of them, as well.
It's the ontology bugbear. Any classification system is somewhat 
arbitrary, culturally determined, overlapping, fuzzy, broken.
Gotta start somewhere though, and keep in mind that the categories 
you're working with are not real or immutable. Recommended starting 
points would be welcome.

>  > At least, that's the closest thing I'm likely to find to what I
>  > *really* want, which is a library of archetypes I can use to design a
>  > bunch of procedural models (that's both 3D models and conceptual
>  > models) for users to populate our VR space with.
>
>To date, the VTP is the only existing implementation of such procedural
>'archetypes' - if you mean such things as roads, trees, and buildings.  I
>would encourage you to build on the work already done.

Never fear, i'm too lazy to do otherwise ;)
I've seen the VTP list here:
http://www.vterrain.org/Culture/culture_class.html#The%20Classification
and yes, I'm using it for the moment, so thank you :)

It's a good start, but needs extension for my purposes.
For example, House could be broken down various ways. I really am 
thinking in terms of a loosely-hierarchical set of attributes; for 
example somthing like:

House:
	Style: Terrace (archetype: Workers Cottage)
		Single-storey
			Period: Victorian
			Built: 1890-1900
		Rear extension
			Period: Modern
			Built: 1950-1970
	Colour: Yellow

With this info you could build a procedural model on a defined block 
that would approximate the appearance of the  feature. From a map 
view, you could choose to represent it with an appropriate sigil; as 
you approached it you could trigger a level-of-detail tree. All this 
could be set up in the client and server so as to allow the 
publication of this information by a non-expert user without 3D 
modelling or geospatial training. Optionally, the lowest level of the 
LOD tree could contain an accurate model of the building. Or if not 
available, even interiors can be modeled procedurally - the 
CHARISMATIC project at U. Surrey did some good work on this.

V.
-- 
Viveka Weiley, Karmanaut.
{ http://www.karmanaut.com | http://www.planet-earth.org
    http://www.MacWeb3D.org | http://sydney.siggraph.org.au }
Hypermedia, virtual worlds, human interface, truth, beauty.