Your fvc.facevertexcdata corresponds to the FaceVertexCData property, and your "one dimensional" specification there corresponds to the "indexed color" case,
An n-by-1 matrix, where n is the number of rows in the Faces property, which specifies one color per face.
In turn, the handling of CData and FaceVertexCData is controlled by the property
CDataMapping
scaled — Transform the color data to span the portion of the colormap indicated by the axes CLim property, linearly mapping data values to colors. See the caxis command for more information on this mapping.
direct — Use the color data as indices directly into the colormap. When not scaled, the data are usually integer values ranging from 1 to length(colormap). MATLAB maps values less than 1 to the first color in the colormap, and values greater than length(colormap) to the last color in the colormap. Values with a decimal portion are fixed to the nearest lower integer.
"scaled" is the default and is what you are using. The lowest value in your FaceVertexCData is being mapped to the lowest color index and the highest value in your FaceVertexCData is being mapped to the highest color index and everything else is being linearly transformed inbetween.
Thus to use a specific color for a particular face, you should find the desired color in the colormap, calculate its position as a fraction of the number of colors in the map, then multiply that fraction by the range of intensities being considered, (max(FaceVertexCData)-min(FaceVertexCData)), and add min(FaceVertexCData) to that. Now set the face's entry in the FaceVertexCData vector to that back-calculated value.
If, however, the desired color is not normally in the colormap, then you have a bit more work to o. You need to calculate the color number that each intensity maps to, replace the FaceVertexCData values by the appropriate color number, and change the CDataMapping property to 'direct':
colormap(jet);
numcol = size(colormap,1);
min_inten = min(FaceVertexCData);
cfrac = (FaceVertexCData - min_inten) ./ (max(FaceVertexCData) - min_inten));
cnum = 1 + floor(numcol * (cfrac * (1-eps)));
and set the FaceVertexCData property to cnum.
Multiplying cfrac by (1-eps) is correcting for a boundary problem. The fraction will be exactly 0 at the minimum intensity, and could potentially be exactly 1 at the maximum intensity (round-off error permitting.) Unless limited precision dictates otherwise, only the very maximum intensity can map to exactly 1, with everything else mapping to at least slightly below 1. If you multiply this by the number of colors you would get a value from exactly 0 to (potentially) the number of colors. But you want 1 to the number of colors. If add 1 then the highest intensity could potentially end up mapping to one more than the number of colors. To avoid this possibility, we multiply the color fraction by just slightly less than 1, so that the maximum value is now slightly less than 1; multiply that by the number of colors and the maximum is now just slightly less than the number of colors, with the minimum mapping to exactly 0. Now take the floor() and you will get 0 to (one less than the number of colors). Add 1 and you get 1 to the number of colors.
Now that we have the vector of color indices into the color map, to set a face to a color outside the color map, we pull a trick: we add another color to the color map, and we set the FaceVertexCData to be the number of the new entry. Because we already switched to 'direct' instead of 'scaled', the entries we already calculated indices for still refer to the same color, but now we can also reference colors that were not there before. For example,
colormap([colormap;0 0 0])
and now the last entry of the colormap is [0 0 0], pure black. You can do the same thing for pinning the color of other faces.
