How to make the marker size in a scatter plot 'to scale' with the axes?

I am using 3D scatter command using the following syntax: scatter3(X,Y,Z,S). The marker size is variable in my case where S is a vector. But the marker size is not in scale with the plot axes. How to make the marker size proportional to the axes units?

[Ahmet Cecen]

Can you take a screenshot with something like snipping tool and use annotation (as simple as paint or powerpoint) to show what you currently have and what you want to see?

Answer 1

[Walter Roberson]

https://www.mathworks.com/help/matlab/ref/scatter.html#inputarg_sz

"The units for the marker area is points squared."

That is the same "points" as specifying 'points' for axes Units property

If you want to make something proportional, then

ax = gca;
old_units = get(ax, 'Units');
set(ax, 'Units', 'points');
pos_points = get(ax, 'Position');
set(ax, 'Units', old_units);
narrower_part_points = min(pos_points(3:4));

Now create something proportional to narrower_part_points. But remember that the size for scatter()'s purpose is area, which is points square. So for example something rand with diameter 1/20th of the width should be expressed as pi * narrower_part_points.^2 / 4

the division by 4 is the change between pi*r^2 and pi*(d/2)^2

Answer 2

[KSSV]

x = rand(10,1) ;   % some random x 
y = rand(10,1) ;  % some random y 
sizes = randsample(1:100,10) ;   % random sizes for 1o points, which lie between 1 to 100
scatter(x,y,sizes,sizes,'filled')  ;  % scatter plot with varying sizes and sizes as color on them

You can extend it to 3D case.

[AS]

Sizes are different for markers but the size is not in the same scale as that of the axes.

[KSSV]

you fix the sizes depending on your scale.

Answer 3

[Will Reeves]

I'd like this functionality too, but I would like the simplicity of being able to pick an "absolute size" drawn in one (or more) of the axis units. For instance I currently use the "viscircles" function, found in the image processing tool box, to achieve this. However, it's really really slow if you want to visualise thousands of cirlces of say 20 micron radius on a plot. It also doesn't offer the "filled" functionality of the scatter plot.