[Paraview] GIS with ParaView - how to drape?

[John Harrop]

Berk,

Thanks for your comments.  I think you understand the goal correctly. 
Here is a little more background and some of our thinking and interests.

On Tue, 2004-03-09 at 07:01, Berk Geveci wrote:
> If I understand it correctly, you want to perform two distinct
> operations:
> 1. modify the z coordinate of each point of a regular grid with an
> elevation (this is what a carpet plot is and as explained in previous
> messages, it is easy to do this in ParaView using "warp")

The elevation model may be in a regular grid - we could always convert
it to that.  I had not been thinking of using ParaView to do this, I had
assumed we would need to build an irregular 3D mesh.  There is some
debate/confusion over whether this is a 3D or 2D object.  In GIS and
geoscience we often call it a 2.5D (thats 2 1/2 D not a version number
;-) surface.  The surface is constrained by the fact it cannot fold back
over itself.  Categorical and continuous attributes are typically
"projected" onto it, but generally the original projected data does not
share nodes with the elevation model.

> 2. apply an image as a texture on a 2D surface (texture mapping).
> Although the underlying engine (VTK) supports this, ParaView does not
> yet support texture maps. Two reasons: 1. there has not been enough
> demand to make this a priority, 2. since ParaView is designed to run in
> parallel (and perform parallel rendering), there are some technical
> issues we haven't addressed.

The visualization demands generated by mineral exploration are not very
big in a business scale.  Consequently, we are presenting two papers at
a conference in Toronto that suggest open source, freely available tools
and much better collaboration with other fields may be the best economic
model for supporting geoscience software in the mineral exploration
industry.  Most GIS does not need full 3D support.

A solution we have been evaluating is to stop expecting a 3D package
that also supports high resolution paper and electronic map creation (2D
world products).  Instead we are building a suite of modular filtering
and inversion tools for processing data, and extending GIS systems such
as GRASS which can support 2d map management and creation.  3D system
generally have difficulty supporting the same resolution we need for
surface mapping (think of publication quality map products).  Being able
to export from the high-res 2D system and view this data in full
resolution on a 2.5D surface (while 3D volumetric data remains at an
appropriate lower resolution) would be a major success.

> Depending on your requirements, you can use ParaView together with VTK
> (VTK supports C++, python, tcl and java development) or you can dig in
> ParaView code (with some help from developers) and try to enable texture
> mapping.
>

We are looking at writing a series of GRASS export utilities to go from
GRASS formats to vtk and thus ParaView or MayaVI.  For now we could
convert to a dense regular grid with attached attribute data.

We would be interested in digging into the code and looking at enabling
texture mapping.  Sounds like it might be a co-op student size project.
 

John Harrop


> -Berk
> 
> On Sun, 2004-03-07 at 13:05, John Harrop wrote:
> > Our group is fairly new to ParaView/vtk but our initial results and
> > impressions have been very favourable.  We work with 3D geophysical
> > models that we build by geophysical inversion of (mostly) surface
> > surveys.  ParaView has worked very well at enabling us to place drill
> > hole traces through volumetric models.  It has also been very good at
> > positioning 2D sections in a 3D model.
> > 
> > One area we have not quite figured out yet is: How do you drape a 2D
> > texture/image onto a surface?
> > 
> > For example, suppose you have a satellite image or air photo of the
> > surface of the area you have been modelling.  We usually also have a DEM
> > so we can generate an elevation surface as a regular mesh or a TIN.  Can
> > anyone make suggestions on where I can learn more about this kinds of
> > draping/texture mapping.  (Incidentally, in this case the image is not
> > being "corrected" by the terrain - its just being dropped vertically
> > onto the surface with no horizontal adjustment of pixel positions.)
> > 
> > Cheers,
> > 
> > John Harrop
> > 
> > 
> > 
> > _______________________________________________
> > ParaView mailing list
> > ParaView at paraview.org
> > http://www.paraview.org/mailman/listinfo/paraview
> -- 
> Berk Geveci <berklist at nycap.rr.com>
> 
> _______________________________________________
> ParaView mailing list
> ParaView at paraview.org
> http://www.paraview.org/mailman/listinfo/paraview