3D Plot with large data

Hi, I have a question regarding doing a 3D plot.

I have a 500,000x3 matrix (column A, B, C). Now I want to do a 3D plot by showing A in x-axis and B in y-axis and C in z-axis. I know I could probably use funtion "mesh" to do the plot. But to use the mesh function, I have to use meshgrid function to transform the vectors into matrices. The problem is that since each variable has 500,000 rows, I got an error message "Maximum variable size allowed by the program is exceeded." Anyone has a solution to it or other ways of doing the 3D plot?

Thanks for your help!

Stephen

[Oleg Komarov]

Don't recall the post, but you won't need all of the points especially since the amount of pixels on the screen are less...

[changiz]

u can use isosurface ,if you matrix name is mat: isosurface(mat);

Answer 1

[Jiro Doke]

Another thing you could try is to do some interpolation of the data and plot at a lower resolution. As Oleg points out, there's no sense in trying to visualize 500000 points at once. Check out triscatteredinterpclass (or griddata).

F = TriScatteredInterp(M(:,1), M(:,2), M(:,3));
[qx, qy] = meshgrid(linspace(min(M(:,1)), max(M(:,1)), 20), ...
                    linspace(min(M(:,2)), max(M(:,2)), 20));
qz = F(qx, qy);
mesh(qx, qy, qz);

[Walter Roberson]

Or gridfit,

http://www.mathworks.com/matlabcentral/fileexchange/8998

Answer 2

[Walter Roberson]

You have not really defined what you want your plot to look like or whether the points are expected to have relationships to adjacent points.

plot3(M(:,1), M(:,2), M(:,3))

scatter3(M(:,1), M(:,2), M(:,3))

[stephen]

Thanks for your quick response, Walter.

Each pair of A and B decides an unique value of C. The plot would like a curved surface. I tried plot3 but looks like the screen froze for a little while then crashed. I guess it was because the data is too big.

[Walter Roberson]

Sounds like you don't have any structuring of the points. In that case, scatter3() would be better. You could start off with a random sampling of points to see how it goes:

numsamps = 1000;

idx = 1 + floor(size(M,2) * rand(numsamps,1));

scatter3(M(idx,1), M(idx,2), M(idx,3));

Answer 3

[changiz]

u can use isosurface ,if you matrix name is mat: isosurface(mat);