[Paraview] FastLIC implementation

Tue Oct 3 10:15:35 EDT 2006

Dear list,

sorry for the long posting to come, but I'd like to introduce to all of
you the result of a MSc. thesis performed here at KTH by Markus
Ringlander, the result is a:

		FASTLIC ALGORITHM CODED IN C++/VTK 
	(... with the goal to be included in ParaView as a filter.)

As opposed to the LEA implementation by Jean Favre, this program can
deal with unstructured (ultimately: triangles), 3D surfaces. For more
details, see:

http://www.ave.kth.se/staff/aero/crippa/lic.html

As of today, the program can't be incorporated in ParaView, due to a
missing texture mapping actor, see:

http://www.ave.kth.se/staff/aero/crippa/lic.html#remaining_issues

Is there somebody at Kitware, or elsewhere that has spent a thought on
how to get this actor into ParaView? 

Just to point out that I/we are not the only ones interested in having
LIC in ParaView, I searched the vtkusers/paraview mailing lists and
found a hit as far back as Dec. 2000 ... where Berk said something about
incorporating LIC implementations made available to him :-) (see below)

Well ... here's our shot:

http://www.ave.kth.se/staff/aero/crippa/info_staff_files/lic-source-1-2.tar.gz

http://www.ave.kth.se/staff/aero/crippa/info_staff_files/lic-source-1-2.zip

Regards, Simone Crippa

=========================================================================
[vtkusers]/[Paraview]

Markus Ringlander; Sat May 27 20:19:15 EDT 2006

Hi vtk-users,

I will do my MSc thesis in scientific visualization, and specifically
the task is to do Line Integral Convolution (LIC) on unstructured
surface grids in 3D. I am supposed to use VTK for this task. After
having studied the online documentation for VTK, and done searches on
this mailing list, I have a few wonderings I hope someone can shed a
light on. Let me try to explain them.

The basic technique I will use for this 3D surface LIC is to generate
a texture for each polygon in the grid. I have seen examples on a few
texture mappings, where one can assign texture coordinates to vertices
of the polygons building up the grid. In the end one uses
actor-SetTexture(...). But all examples I have seen refer to one
sample image/texture as source.

In my case, I am going to traverse the elements in the grid, and
generate a 2D texture/image for each element. So I will have many
separate textures.

A bit simplified, I was thinking something like this:

1) Read my unstructured grid from a file.
2) Iterate through each polygon in the grid.
3) For each polygon in the iteration, perform the LIC algorithm on it,
which here gives a texture as output for each polygon.
4) Map texture coordinates from this specific image/texture to the
corresponding real world polygon.
5) Back to 3)
6) Output the whole textured grid on the screen.

As I stated earlier, I have only seen examples of a single actor for
texturing, refering to a single image/texture file. In my case I am
wondering if I need to set up an actor for each texture I create? It
can then be over 10,000 textures...

Please help me with some information wheather this is doable in VTK,
and if so, any helpful code snippet for the assignment of the textures
to each polygon.

Thanks!

/Mark

==========================================================================

Jean M. Favre; Mon Sep 13 02:39:01 EDT 2004

Christpher Auer wrote:
> Hello everyone!
> 
> can someone tell me please, if there is currently some development on
> the implementation of "LIC Algorithm" and/or "Unstructured grid volume
> rendering"? Or do you know someone who might know this?
> 

We have an implementation of the Lagrangian Eulerian Advection algorithm 
published at IEEE Vis 2001. It produces LIC images for steady fields and 
time-dependent LIC-like textures with spatial and temporal continuity 
for un-steady fields.

Source code, Makefile and test routines are provided for VTK and ParaView.

http://www.cscs.ch/~jfavre/Projects/vtkLEA/vtklea.htm

==========================================================================

Thomas Brandner; Sun Jan 9 05:53:52 EST 2000

Dear  vtk users,

I`m currently working on a study project about line integral convolution
(LIC). 
This is a texture-based technique for effectively displaying 2D and 3D
vector fields.

Has anyone managed to implement this type of filter in vtk ?

As it is to be used  in VR-applications, the focus  is not so much on
accuracy but rather on efficiency to maintain interactivity. 

Thanks very much in advance for any help,
Thomas Brandner

==========================================================================

And as far back as Mon Dec 11 10:16:19 EST 2000, Berk said:

LIC, FLIC and spot noise are in my to-do list. I don't know when I can
get to them though. If someone has implementations, I will be happy to
incorporate them to vtk.

	Berk

"David P. Munich" wrote:
> 
> Does anyone know of an available implementation of Line Integral
> Convolution(LIC) in VTK?
> 
> Thanks,
> Dave
> 
> _______________________________________________
> This is the private VTK discussion list.
> Please keep messages on-topic. Check the FAQ at: <http://public.kitware.com/cgi-bin/vtkfaq>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/vtkusers
-- 
Simone Crippa
Research assistant, Dipl.-Ing.
Royal Institute of Technology (KTH)
Aeronautical and Vehicle Engineering
Division of Aerodynamics
Teknikringen 8
SE-10044 Stockholm, Sweden
+46 (0)8 790 6465