[GRASS-dev] Google Summer of Code 2008

why not to use the python bindings of VTK (http://www.vtk.org) to do 3D
rendering? It seems to me that VTK is a bit easier to handle than OpenGL
directly, but my programming skills are not the best.
WolfgangZ

Apart from the problem with limited coordinate ranges brought about by
OpenGL, VTK uses a visualization pipeline rather than layers. There is
nothing wrong with this per se, but GIS users are accustomed to layers
that cover each other in a pre-defined order, even if there is no
difference in the Z coordinates. VTK projects all objects into the same
3D space, potentially giving unexpected results, e.g. when overlaying
vector outlines on a raster surface without raising the former a little
bit towards the zenith.

Might be worth thinking about this. I don't know how easy or complicated
it would be to emulate a layer-based drawing on a VTK rendering
pipeline.

Benjamin

VTK supports 3d render windows as well as 2d render windows and 2d actors/mappers to visualize
image data. So you are able to use several 2d actors to create layer based visualizations.

http://www.vtk.org/doc/nightly/html/classvtkActor2D.html

If we are able to bring raster, volume and vector into VTK, we will be able to use the hundreds of
VTK classes (image, volume and vector) to process grass data.

http://www.vtk.org/doc/nightly/html/classes.html

Which IMHO will bring us enormous benefit in form of parallel image processing and
complex image-vector, vector-vector and vector-volume algorithms.

Just my 2c
Soeren

2008/2/25, benjamin.ducke@ufg.uni-kiel.de <benjamin.ducke@ufg.uni-kiel.de>:

why not to use the python bindings of VTK (http://www.vtk.org) to do 3D
rendering? It seems to me that VTK is a bit easier to handle than OpenGL
directly, but my programming skills are not the best.
WolfgangZ

Apart from the problem with limited coordinate ranges brought about by
OpenGL, VTK uses a visualization pipeline rather than layers. There is
nothing wrong with this per se, but GIS users are accustomed to layers
that cover each other in a pre-defined order, even if there is no
difference in the Z coordinates. VTK projects all objects into the same
3D space, potentially giving unexpected results, e.g. when overlaying
vector outlines on a raster surface without raising the former a little
bit towards the zenith.

Might be worth thinking about this. I don’t know how easy or complicated
it would be to emulate a layer-based drawing on a VTK rendering
pipeline.

Benjamin


grass-dev mailing list
grass-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-dev