Seeking an efficient method for using 'scatter3' to create a 3D scatter plot of fixed receiver positions.

Hi,

I wish to represent a 3D array of reciever positions as dots - see attached plot for the general idea (I won't share the shameful code used to generate it, haha). My intention is to then animate the plot by representing each position's signal amplitude by its corresponding dot's colour (each sample generates a new plot/frame). I've successfully animated plots of a 2D plane of receiever positions, with amplitude on Z-axis, so I'll cross any issues with that bridge once I've had a good attempt myself, but I welcome any quick tips or relevant docs people may have about sending the time dimension data to the figure's colour info.

My only question at this stage is actually about how to most effieciently plot a 3D "block" of dots (as below) according to specified dimensions - basically - in very broad terms - to input 3 x 3 x 3 and get something resembling the attached image (or any three dimensions for that matter). The solution is almost guaranteed to be embarassingly simple, but it has eluded me so far (I've been ill, so my concentration is totally shot, which hasn't helped).

Many thanks in advance for any advice, tips, links or hints, etc.

Cheers!

Answer 1

[Max Heiken]

I think you are looking for meshgrid.

[y, x, z] = meshgrid(1:3, 1:3, 1:3);
scatter3(x(:), y(:), z(:), [], amplitude(:), 'o', 'filled');

Notice that it is [y, x, z] and not [x, y, z], it is a quirk of the meshgrid function.

[Max Heiken]

Assuming the amplitude is stored with time in the 4th dimension

amplitude = rand(3,3,3,n);

then the animation (in the same figure in this case) could be achieved by something like

h=scatter3(x(:),y(:),z(:),[],reshape(amplitude(:,:,:,1),1,[]),'o','filled');

for epoch=2:size(amplitude,4)
    set(h, 'CData', reshape(amplitude(:,:,:,epoch),1,[]));
    pause(0.1)
end

[Peter Beringer]

Thank you very much for such an elegant solution. The very obvious thing that was eluding me and causing puzzling errors was that I was not calling the output for 'z' as well as 'x' and 'y' when using 'meshgrid' ... oh, dear, stuck in Edwin Abbott's "Flatland". ;)

In reality, the recordings made at receiver positions will have time/amplitude in the first dimension, but that would only involve a simple permute. Measurements are going to be taken by a linear reciever array moving in lateral increments at each vertical increment, with the recording repeated at each position, so the data will also need concatenating into matrices.

Many thanks again for the help. It might make quite an interesting plot - not unlike early radio telescope array data. Haha.

Cheers!

Answer 2

[Chunru]

Try scatter3:

t = 0:pi/5:4*pi;
xt = sin(t);
yt = cos(t);
h = scatter3(xt,yt,t,40, t*30, 'MarkerFaceColor', 'b');
colormap(hsv(512))
colorbar
box on
grid on
caxis([0 512])

% Animation
for i=1:20
    h.CData = rand(size(t)) * 512;
    drawnow
    pause(0.1);
end

[Max Heiken]

OP already knows about scatter3 as we can infer from the attached image.

[Chunru]

The OP plot can be produced by plot3 as well :-)

[Peter Beringer]

This is another great solution. Might have to use 'tic'/'toc' and race the two. ;)

Many thanks.