Why does my surface have a 'jagged' look when I do this?

Hi guys,

I am trying to do this: https://www.mathworks.com/matlabcentral/answers/214353-simple-question-how-to-graph-certain-surfaces-depending-on-the-z-value

I think I have successfully done this. However, I would like to know the reason why I get a 'jagged' appearance on the bottom surface? and is there a way to make it smoother? Here's the code:

x = [0:100];
y = [0:100];

Test1 = @(x,y)(x+y);
Test2 = @(x,y)(x.^2+y);
[X1,Y1] = meshgrid(x,y);
Z1 = Test1(X1,Y1);
Z2 = Test2(X1,Y1);

i = Z1<50;
Z1(i)=NaN;
s1 = surf(X1,Y1,Z1);
hold on
i = Z2>=50;
Z2(i)=NaN;
s2 = surf(X1,Y1,Z2);

Thanks

Answer 1

[Joseph Cheng]

It is due to the spacing of your points. the data is being graphed in a grid so you're not going to have a nice straight line connecting the diagonals of the grid. you can reduce the jagged appearance by decreasing the spacing of the points. for instance if you try running your code with

x = [0:.5:100];
y = [0:.5:100];

you'll see that the jaggedness is decreased

[A]

Thanks very much. That makes sense. However, when I do that, the figure goes 'very' black. I think it's probably because of the lines? Is there a way to adjust the way lines are drawn so that you only see a line at intervals of 1?

[Brendan Hamm]

No, but you can turn the lines off.

surf(X1,Y1,Z1,'LineStyle','none')

Answer 2

[pfb]

Hi

Of course, making the grid thicker reduces the jaggedness.

As far as you are plotting planes, you can easily obtain nicer results.

Since these are planar manifolds, you can use one large patch instead of many small patches as in surf.

For that, look into "patch" or "fill3".

If you insist on using surf, you can use a grid adapted to your plane. That is: one of the directions should be the projection of the isolines, the other orthogonal to that (greatest gradient). You can see that when your plane depends only on X or Y. That should work also on your X.^2+Y, although you might want to use different spacings along the two lines.

Finally, you can perhaps use a triangulation for the original mesh. Look at "trisurf" and "delaunay".

Update I just realized you need to use a slightly different strategy for that.

For instance

x = [0:100];
y = [0:100];

Test1 = @(x,y)(x+y);
Test2 = @(x,y)(x.^2+y);
[X1,Y1] = meshgrid(x,y);
Z2 = Test2(X1,Y1);
i = Z2<=50;
X1=X1(i);
Y1=Y1(i);
Z2=Z2(i);
tri=delaunay(X1,Y1);
trisurf(tri,X1,Y1,Z2);

This eliminates the jaggedness. Also, I think you do not need a very thick grid for a nice result.