RE: ElevationGrid

["Crispen, Bob" <Robert.Crispen@HSV.Boeing.com>]

Don Casteel set my mind to ticking with:

> Point taken, however if a widly accepted and successful way of doing
> things
> exists, why not use it. An image file is already a rectangular grid,
> each pixel
> contains numerical information. They are compact and well known, and
> there
> are probably more tools available for working with image formats than
> text formats.
> 
> In addition, think of what you could do with an animated GIF if it was
> directly
> processed as a height field.
> 
> As for authoring tools............ what's an authoring tool?    :-)
> 
Well, I've been thinking of a cute little tool called
Terrain Forge by Silver Software:
http://www.geocities.com/SiliconValley/Park/7425/
It was a lot of fun to go into a paint program like
Paint Shop Pro or something, use the airbrush,
and make some terrain.

Alas, last I looked the best way to get VRML was
to save as DXF and then use Crossroads or
something to convert to VRML IndexedFaceSets.

Which led to a number of questions:

(1) Do folks in the geo community actually use
height field maps, or is this something that's
pretty much in the domain of modelers and
raytracers?  Is there a file format that's used
a lot to represent height fields?

(2) Converting between IndexedFaceSets and
ElevationGrids should be a piece of cake.  All
I need to know is the best way to interpolate
and the best way to set the grid spacing and
I'm there.  What are the tricks folks who do
this for a living do to convert back and forth
between regularly spaced elevation data and,
what would you call it?, feature-based elevation
data?  Or don't you have anything equivalent
to the latter?
---
In case you don't know, there's a good
discussion going on on www-vrml (which I
think started here but ended up over there)
about double precision coordinates.  With
any luck at all, we'll get some metrics on
the speed hit you take when you use doubles.

That will certainly help us focus on whether
we'll need a special set of nodes for doubles
or whether we can let browsers (through a
switch?) use doubles for all floats.

It might also help if we could do some analysis
on where and when we really, truly need double
precision.  What's the domain for which
double precision is a hard requirement?  That
might lead us to some solutions that are a
bit more like scalpels than hammers.
---
And how *is* color data specified?  I would
have thought satellite photos would do the
trick, and (in VRML) you'd simply tweak the
image for satellite angle and Earth curvature
and apply the image as an ImageTexture to
the ElevationGrid or IndexedFaceSet.  What
am I missing?
---
Since my questions (and probably a number
of questions from folks who know VRML but
not geo stuff) seem so elementary, is there
some set of websites or books that can give
us an elementary overview so we can at least
start talking the same language?

I'm really committed to doing what it takes
to make VRML a useful tool for this community
and I'm basically trying to get a toehold.

Bob Crispen
bob.crispen@boeing.com