If you have a surface in Visit (ie an isosurface, or a slice), you can save the window as a vtk file. And then use the attached read2DScalarVTK.m matlab function to import the vtk file as a struct in matlab. The struct returns contains

• Faces - Fx3 array of integers containing the point numbers that make up each face
• Vertices - Vx3 array of floats containing the xyz coordinates of each vertex
• FaceVertexCData - Fx1 scalar field defined on the faces

Example

The following matlab script will prompt for a file, import the data, and plot the surface in matlab using the default colormap

[filename,pathname]=uigetfile('*.vtk','Please select a file','*.vtk')
data=read2DScalarVTK([pathname,filename])
patch(data,'LineStyle','None','FaceColor','flat')

Line integral convolution

You can also use velocity data to do line integral convolution

% Fix these to point to the LIC toolbox and update the pathname to your rho.vtk, vx.vtk, and vz.vtk files
addpath('T:\Documents and Settings\jcarroll\My Documents\MATLAB\LIC\toolbox_image\')
addpath('T:\Documents and Settings\jcarroll\My Documents\MATLAB\LIC\toolbox_image\toolbox\')
pathname='Y:\visualizations\iigu\2\data\';

%% Read in rho.vtk, vx.vtk, and vz.vtk (since my slice was in the y direction)
N=512; %resolution of lic image

filename='rho.vtk';
data=read2DScalarVTK([pathname,filename])
[xx,yy]=generatemesh(data,N); %We are going to use the same mesh for all regridding all of the data
rho=regridmesh(data,xx,yy);

filename='vx.vtk';
data=read2DScalarVTK([pathname,filename]);
vx=regridmesh(data,xx,yy);

filename='vz.vtk';
data=read2DScalarVTK([pathname,filename]);
vz=regridmesh(data,xx,yy);

%% Normalize velocities since this is expected by LIC routine
maxnorm=sqrt(max(max(vx.^2+vz.^2)));
vx=vx/maxnorm;
vz=vz/maxnorm;

%% Set the options and perform the line integral convolution
options.histogram = 'linear';
options.verb = 0;
options.dt = .5; % time stepping for LIC
options.flow_correction = 1;
options.niter_lic = 10; % several iterations gives better results
options.M0=randn(N); %mask for noise
z=zeros(N,N,2);
z(:,:,1)=vz;
z(:,:,2)=vx;
result=perform_lic(z,10,options);
imageplot(result)

%% Now we can use the noise mask generated from LIC on the velocity data, to adjust the brightness of a false color image of the density
f=.8; %Brightening factor - so that you can still see color through black/white noise
rho_min=min(min(rho)) %get lower bound
rho_max=max(max(rho)) %get upper bound
cmap=colormap('hsv');
cmap=cmap(1:200,:); %restrict colormap to first 200 entries instead of 256 to avoid circling back around to red
cdata=interp1(linspace(log(rho_min),log(rho_max),size(cmap,1)), double(cmap), log(rho(:))); %get cdata from scalar field using log scaling
cdata=reshape(cdata,[N,N,3]).*repmat((1-f)+f*result,[1,1,3]); %multiply by mask
image(cdata) %draw image