Matlab: `mesh()` plot with less number of grid

Question

Let's say data is a matrix of size 129 * 129.

by using

mesh(data, 'FaceColor', 'none', 'EdgeColor', 'black')

we get something like

We can find that the grid are quite intense. I would like to have the same figure but with less number of mesh lines, something like

It is of course doable to plot a smaller data, for example data(1:10:end, 1:10:end). But in this way, the plot are not accurate as before anymore.

Another example is plot(..., 'MarkerIndices', ...). This can give you a plot with less number of markers without modifying the plot. https://www.mathworks.com/help/matlab/creating_plots/create-line-plot-with-markers.html

So you want less lines without drawing less lines? I have some troubles understanding your goal. — Daniel, Jan 21 '20 at 20:03
@Daniel yes, that's my goal. For example by using `plot(..., 'MarkerIndices', ...)` you can have less makers without drawing less points. — null, Jan 21 '20 at 20:06
Do you perhaps want a [`surf`](https://www.mathworks.com/help/matlab/ref/surf.html) plot with *no* grid lines? — gnovice, Jan 21 '20 at 20:06
@gnovice yes. maybe we could `surf` plot first and then `mesh` with less lines? — null, Jan 21 '20 at 20:07
@NathanExplosion: with MarkerIndices you only reduce the number of markers. With your mesh you are already at 0 markers. I guess you have to reduce the amount of data displayed. — Daniel, Jan 21 '20 at 20:44

gnovice · Answer 1 · 2020-01-21T20:18:56.690

I think your best option would be to create a surf plot with no grid lines (showing a colored surface with the full resolution of your data), then overlay a down-sampled mesh plot. Something like this:

surf(data, 'EdgeColor', 'none');
hold on;
mesh(data(1:10:end, 1:10:end), 'EdgeColor', 'black');

You could also add some transparency to the surf plot to make the mesh visible through it:

surf(data, 'FaceAlpha', 0.7, 'EdgeColor', 'none');

score 3 · Accepted Answer · answered Jan 21 '20 at 23:37

3

An alternative approach is to use plot3 to plot the mesh lines manually. That way you can plot each line smoothly using all the data points, but not have as many lines.

[X,Y,Z] = peaks(201);
step = 5;

plot3(X(:,1:step:end),Y(:,1:step:end),Z(:,1:step:end),'k')
hold on
plot3(X(1:step:end,:).',Y(1:step:end,:).',Z(1:step:end,:).','k')
hold off

answered Jan 21 '20 at 23:37

David

8,449
1
22
32

Nice, I think this is a better option than just downsampling data in a mesh. – gnovice Jan 22 '20 at 02:12

null · Answer 3 · 2020-01-22T11:56:52.617

The answer from @David is good. In addition to his approach, we can also replace plot3 with many infinitesimal mesh. The idea is to plot mesh for single vectors many times.

[X,Y,Z] = peaks(201);

tempz = NaN(201, 201);

tempz(1, :) = Z(1, :);
mesh(X, Y, tempz, 'EdgeColor', 'interp');
hold on

% plot x lines
for i = 2:10:201
    tempz = NaN(201, 201);
    tempz(i, :) = Z(i, :);
    mesh(X, Y, tempz, 'EdgeColor', 'interp');
end

% plot y lines
for i = 2:10:201
    tempz = NaN(201, 201);
    tempz(:, i) = Z(:, i);
    mesh(X, Y, tempz, 'EdgeColor', 'interp');
end

The original is

By using the snippet above, it gives

The benefits of this over @David's answer is that you can preserve all of the fancy properties of mesh, for example shading interp etc.

Matlab: `mesh()` plot with less number of grid

3 Answers3