cellox

03-07-2003, 08:26 AM

Hi everyone,

This is a bit interdisciplinary field but i cant help maself to ask this qtn. I have

developed a Finite Element Analysis tool which makes stress and displacement computations

for structures composed of 2D quadrilateral elements. Any unknown value is computed

~exactly on the

nodes, say corners of a quadrilateral and for the rest of the domain it is interpolated

makin use of some shape functions.

I guess it is clear up to this point.

While plotting the colour contours, i divide my main element into a number of rectangles

(or polygons) for whose corners i compute the unknowns using my shape functions, and give a

color code to every corner then i draw that solid polygon, which makes the program use the

linear color interpolation of OpenGL itself.

Now the problem is:

I think that this is such an expensive method of doing this job. I'd be pleased if someone

shows me a better, faster and more efficient way of doing this job.By my method because of

such heavy computations pan, zoom, etc options of the interface really sucks for large

systems having say more than 3000 nodes...

I am in trouble!!! http://www.opengl.org/discussion_boards/ubb/smile.gif

This is a bit interdisciplinary field but i cant help maself to ask this qtn. I have

developed a Finite Element Analysis tool which makes stress and displacement computations

for structures composed of 2D quadrilateral elements. Any unknown value is computed

~exactly on the

nodes, say corners of a quadrilateral and for the rest of the domain it is interpolated

makin use of some shape functions.

I guess it is clear up to this point.

While plotting the colour contours, i divide my main element into a number of rectangles

(or polygons) for whose corners i compute the unknowns using my shape functions, and give a

color code to every corner then i draw that solid polygon, which makes the program use the

linear color interpolation of OpenGL itself.

Now the problem is:

I think that this is such an expensive method of doing this job. I'd be pleased if someone

shows me a better, faster and more efficient way of doing this job.By my method because of

such heavy computations pan, zoom, etc options of the interface really sucks for large

systems having say more than 3000 nodes...

I am in trouble!!! http://www.opengl.org/discussion_boards/ubb/smile.gif