Understanding Meshgrid, interp3 and scatter3

I have some 3d-simulation data I want to process in matlab using the scatter3 function, but despite reading all the documentation and looking for tips I'm stuck.

Current code:

load('TE_period_150nm_testres.mat');

Ex=Monitors.model_volumepower.Ex;
Ey=Monitors.model_volumepower.Ey;
Ez=Monitors.model_volumepower.Ez;
Ex=abs(Ex);
Ey=abs(Ey);
Ez=abs(Ez);
mag=sqrt(Ex.^2+Ey.^2+Ez.^2);

x=Monitors.model_volumepower.x;
y=Monitors.model_volumepower.y;
z=Monitors.model_volumepower.z;

x2 = linspace(min(x),max(x),max([length(x) length(y) length(z)]));
y2 = linspace(min(y),max(y),max([length(x) length(y) length(z)]));
z2 = linspace(min(z),max(z),max([length(x) length(y) length(z)]));

[X,Y,Z]   = meshgrid(z,y,x);
[X2,Y2,Z2] = meshgrid(z2,y2,x2);

magnew = interp3(X,Y,Z,mag,X2,Y2,Z2);

Mag is a 61*74*53 matrix, x y and z are vectors of size 61, 74 and 53 respectively. I understand Mag needs to become an M*M*M-matrix to work for scatter3. I tried to make a new 74*74*74-matrix (spanning the same range) and use interp3 to interpolate required values. For some reason, all my magnew-values become NaN.

Who can help me out?

(If I can fix this, I'll still have four 3d-matrices (for x, y, z and magnitude) whereas scatter3 apparently needs one long vector. Can I just fix that using

for i=1:length(magnew).^3
magfinal(i)=magnew(i)
end

where I'm basically making it one long vector? (Same for x, y and z)

Thanks a lot in advance!

[Chad Gilbert]

Generally, you would use scatter3 to do something like e.g. show where a set of data points are located in 3D. So it might be interesting to look at something like:

    scatter3(x, y, z);

But you've interpolated a scalar value onto a regular 3D grid. It might be interesting to look at something like (untested):

   scatter3(X(:),Y(:),Z(:),mag(:));

which will use the color of each scatter point to show you the color, but I would expect this to be quite ugly and difficult to make sense of. It's generally very hard to visualize a scalar field in 3D on a 2D monitor.

If it were me, I'd try plotting several 2D slices through the graph using pcolor.