Re: Pb with structured array with irregular distance between points

New Message Reply Date view Thread view Subject view Author view

From: Ken Schwarz (kschwarz@schwarzsgi.dallas.sgi.com)
Date: 06/16/2000 13:04:40

I understand your reasoning for not wanting to comprimise on
re-interpolating your data. I'm speaking strictly from a performance
perspective. Since there is not an OpenGL construct for performing a
non-linear scale of polygonal data, as I see it you have three choices:

1) Pre-process the data
2) Visualize the data as a "distorted" volume
3) Render the volume with your own routine to accommodate for the
    non-linear distortion

Options 1 and 2 will always produce faster renderings.

Best,

Ken

  ...... Ken Schwarz Tech Consultant (TC) SGI Dallas kschwarz@sgi.com ...
... 972.628.3232 Vmail // 888.986.8243 Pager // 972.672.7278 Mobile ...
... http://reality.sgi.com/kschwarz_dallas ...
 ... Brain: Pinky, what rhymes with June? Pinky: No it doesn't. .....

On Fri, 16 Jun 2000, Jean-Christophe PENALVA wrote:

> Hello,
>
> First, it will be difficult to do power of 2 number of points
> (there
> are already "to much points", we don't like to add some !).
> When i speak about the array (these are data from CFD
> computation),
> it's a "regular array". There's the same number of points from the start
> to the end (in x y and z). But the distance between points is not
> regular. For example, there's big distance in the area where there's
> less interaction, and there's more points near the rotor (where there
> are lot of interactions).
> I've look in /usr/share/Volumizer.../OpenGL. there are good
> examples,
> but the voglUnstructured is not for my data, there are not unstructured,
> there's not a regular space between the x y and z. It's a rectangular
> array.
> We don't want to interpolate us array to a regular array,
> because it's
> a loss of information.
> In the volumizer documentation, on page 3, SGI speak about
> "Unstructured meshes that are common in CFD analysis.", but us data are
> not unstructured !!
> Thank you.
> --
> ====================================================================
> C.I.N.E.S. Centre Informatique National de l'Enseignement Superieur
> Montpellier FRANCE Tel: 33 4 67 14 14 14 (Fax: 33 4 67 52 37 63)
> http://www.cines.fr
> mailto:jean-christophe.penalva@cines.fr
> =================================================================
>

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Fri Jun 16 2000 - 13:28:17 PDT