[GRASS5] re: Grass GUI

[Soeren Gebbert]

Dear developers and users,
the GUI discussion is getting more and more interesting. I didnt read 
every mail, but i agree with Syd.

Why not using an external library like VTK for visualisation?
Im working with VTK for almost a year and this pice of software is 
awesome!
VTK has almost all the functionality to display every kind of data from 
GRASS.

It supports any kind of vector (2D, 3D), image (2D pixel and 3D voxel), 
polygonal and unstructered grid data,
that can be produced by GRASS and much more (vectors and tensors). Nviz is 
a great pice of software, but VTK provide much more
possibilities to present the data. And you can analyse the data with VTK 
too!
VTK supports multiple render windows for 2D or 3D data, or mixed. It 
supports picking for points and cells,
it provide widgets for cutplanes, scalarbars, it has a rastercalculator, 
sophisticated colortables, delauny triangulation in 2D and 3D, marching 
cubes
algortihm for 3D Imagedata (Voxeldata in GRASS) and so on, it has more 
than 100 filters. You can parallelize it with MPI,
it is "demand driven" and has a great pipeline concept. It is already in 
use within a GIS (MapInfo).
VTK is availible for almost every Unix-system, Windows and MacOS, dont 
matter if hardware or software opengl!
And the BEST, it is very well documented, easy to extend , has a huge 
community and its open source (BSD like license)!

I have written two modules in GRASS 6.1 cvs to export 2D and 3D rasterdata 
into the VTK dataformat (r.out.vtk and r3.out.vtk).
To try out the capacity of VTK you can export your GRASS data and 
visualize it with Paraview or Mayavi.

VTK provides a interface to QT and it works very well. This means, there 
are possibilities to extend qgis with VTK features.

I think the best approach for a sophisticated GIS/visualisation system is 
to combine GRASS with VTK and QT/qgis.
And im absolutely sure nobody will regret the decision to develop the new 
visualisation system for grass with VTK and QT/qgis.
(my company is very happy with this coice :D) Mixing of 2D and 3D data in 
the same window are no Problem at all with VTK.

I will definitely support the development of the new 
GRASS-GUI/Visualisation concept if the developers choose VTK/QT.
I can design the visualisation pipelines, the grass-vtk interfaces, i can 
give support in VTK issues and so on.
In my company im developing a visualisation system for higher order FEM 
with VTK and QT, so i have some kind of experience.
But i have no time to be a full time developer, because i need to finish 
my studies too.

Best regards
Soeren Gebbert


On Wednesday 16 November 2005 19:30, Syd Visser wrote:
> Grass is one of the few GIS systems that is 3D not only 3D raster but
> also 3D vector yet almost all of the interfaces being discussed or
> available such as Qgis, Thuban, Openve, Udig etc appear to be totally 2D
> (some people like to think 2.5D is 3D). The only one that is actually 3D
> is NVIZ and the last time tried to use it it was very good at 2.5D but
> the 3D part was lacking.
>
> Although a display is not really a part of a true GIS it is likely the
> most important part of a GIS in our line of work (geophysics). We have
> to be able to display our end product graphically to the client and
> almost all of our work is now in 3D since as a geophysicist we usually
> try and determine what is below the ground.
>
> We are now using Paraview (paraview.org)and Mayavi
> (mayavi.sourgeforge.net) for our display and again moving away from
> grass. Both of these programs are based on the VTK graphics libraries.
> We are also looking at TVTK from enthough
> (https://www.enthought.com/enthought/wiki/TVTK) mabey a Python wrapper
> for grass using packages such as these. This could also greatly extend
> the scientific part of grass.
>
>
> So my main points are
>
> 1) Why is a 3D GIS even looking at 2D graphics (usually 3D packages can
> aslo do 2D or it can easily be added)
>
>    2.
>
>       There are 3D graphics libraries available like VTK, OpenDX can
>       they not be used. What does NVIZ use is it totally based on 
OpenGL?
>
>    3.
>
>       A GUI appears to me to be jest a wrapper around good libraries so
>       is what GUI to use that important with the exception that it
>       should be myltiplatform.?
>
> To me a big mistake I often see programmers make is they write software
> to work with todays or even yesterdays hardware/software. Remember you
> will never get your software working till tomorrows hardware is 
available.
>
> In other words, unless there is something new out there, i do not see
> how to do 3D graphics without OpenGL. I have not seen a graphics card
> for years that does not a support it. Even most cheap cards with onboard
> video support OpenGL now.
>
> The people at Grass are doing a great job and it has a real unique
> chance to get ahead of most major GIS systems think 3D all of the way
> don't stop half way. I think Nviz is doing a great job but we can do our
> work in linux which I prefer but we require a viewer on the windows
> platforms sorry but that is where our clients are.
>
> Most of our clients do no care how we get there they only want a pretty
> picture or now 3D movie of the end product. So without good graphics and
> a GUI grass will forever linger in academia well mabey that is what the
> majority of people want?
>
>
> Syd
>