Re: ElavationGrid (was: Height maps)

steve ([email protected])
Wed, 10 May 1995 10:05:36 GMT0BST

> > > ElevationGrid {
> > > fields [ SoSFLong numRow, SoSFLong numColumn,
> > > SoSFFloat rowStart, SoSFFloat columnStart,
> > > SoSFFloat rowDelta, SoSFFloat columnDelta, SoMFFloat
> > elevations
> > > ]
> > > numRow = 0
> > > numColumn=0
> > > rowStart=0
> > > columnStart=0
> > > rowDelta=1
> > > columnDelta=1
> > > elevations -1
>
> Just to make sure I understand this. It is like a quadralateral mesh in
> which
> all of the quads are the same size rectangle and you just supply an array of
> scalar height values, right? If placed in object space, it would start at
> x = rowStart, y = yStart, z = elevations[0] and extend in the positive X & Y
> directions.

Correct - actually, how it is meshed should be left to the browser -
generally it would be triangulated as it is impossible to construct
such a height map from planar quads. By skipping over grid points,
browsers could implement some crude form of level-of-detail by
effectively under-sampling the grid.

> Texture Binding would be to the elevation points.

Yes. Material binding would also be to the elevation points. a
PER_VERTEX binding allows colors to be cooked into the mesh (colors
calculated according to elevation, for example) which can the acts as
a base color for textures applied.

>>-------------------------------------<<
Steve Ghee ( [email protected] )
Director of Technology

Division Ltd
19 Apex Court
Woodlands
Almondsbury
Bristol, UK
BS12 4JT
Tel : +44 1454 615554
Fax : +44 1454 615532
>>-------------------------------------<<