MATLAB: 2D contour in 3D

3d visualizationcontourcontourfplotting

I have a visualization question. I am plotting a slice of data as a 2-dimensional contour using contourf(). What I would like to do is rotate this to show where it sits in 3D, and be able to use the 3D-rotation cursor. The problem is, contourf() really thinks this is just X and Y, and the cursor doesn't want to flip things on it's side. Is there a way to take the contour and plot it in a larger 3D context?

Best Answer

contour() and contourf() work oddly. They call an internal routine that creates patches in 3D at the height of the contour planes, and then they run through and set the Z values to 0 deliberately.

An undocumented trick to this process is that for each patch, the Userdata field of the patch is set to a value which is the Z coordinate. This allows you to post-process, run through the patches, fetch their Userdata, and set the Z values back.

Related Solutions

MATLAB: Strange patch behaviour when hgtransform applied to object

I think this is probably a bug in how the ZBuffer renderer does its lighting computations. If you change the figure's renderer to OpenGL, it should look fine:

set(gcf, 'Renderer', 'OpenGL')

One caveat: the OpenGL renderer does not support Phong lighting. So you will notice that even though the patch's FaceLighting property is set to 'phong' it will fall back to Gouraud-style shading when you change renderers.

MATLAB: Contour grid points

contour() and contourf() are almost identical internally. They return a single hggroup object, and that hggroup object have a bunch of Children. Each of the Children is a patch object for a different contour level, one patch object per level, with embedded NaN in the coordinates where needed to jump from one place to another in the drawing.

The code that builds the contours sets the Z coordinates to the level of the contour heights, but then one of the last things that contour / contourf does is reset those coordinates to 0. Fortunately, each patch object has a UserData property that is used to store the contour level that it was for.

Soooo...

chand = contour(....);
cpatches = get(chand, 'Children');
for K = 1:length(cpatches)
  xcoord{K} = get(cpatches(K),'XData');
  ycoord{K} = get(cpatches(K),'YData');
  zcoord{K} = get(cpatches(K),'UserData');
end

Warning: for any contour that is moderately complex, the list of coordinates will be pretty messy, lots of ins-and-outs.

Best Answer

Related Solutions

MATLAB: Strange patch behaviour when hgtransform applied to object

MATLAB: Contour grid points

Related Question