How to animate line on polar axes plots

I'm trying to animate a plot with polar coordinates, ideally to trace a line following a marker. I've tried using the animatedline object but it doesn't seem to take polar coordinates. This is the code I've tried

figure
pax = polaraxes;
pax.RLim = [0, 6];
pax.ThetaDir = 'clockwise';
animatedline(pax, theta, r);

But get the error:

Error using animatedline
AnimatedLine cannot be a child of PolarAxes.

I've also tried iterating with a for loop and polarplot with a "hold on" but it seems super choppy. Any ideas is appreciated.

Answer 1

[dpb]

polar coordinates are "red-haired stepchildren" in Matlab, for sure.

I've not played with the new(ish) PolarAxes, but I think if you simply first create a line handle with the initial point, you can then update the ThetaData, Rdata properties of the line directly. There's an example of a marker following a line at Trace-marker-along-line for a single point; to draw a line you simply add data points to the internal arrays; remember to preallocate an array of NaN for performance instead of dynamically reallocationg by concatenation.

The newer documentation links make the above hard to find; if didn't know it had existed in earlier versions, not sure one would ever find it. I had to do a search for "Animation" to uncover the link. The only thing at the top level I did find took me to the animatedline which we know doesn't work completely universally yet.

NB: EDITORIAL COMMENT/OPINION

TMW gets so enamored with all their new toys they often seem to introduce them before they're fully ripe which is OK if they don't sever the documentation links to the tried-and-true at least until all the bases are covered.

[Mark Duarte]

Holy. Cow. This has been quite the beast to tame. I think I finally have the minimum of what I was trying to do, animate a line being drawn on a polarplot where I defined the axis and view beforehand. Before when I created a pax = polaraxis with my RLim and ThetaDir, then tried to polarplot the data, it would create a new axis on the figure on top of the one I tried to create. Trying to feed pax to polarplot with the data just wasn't working. Also, calling cla to clear axes would create a default Cartesian axis that would show up in the figure too!

I had to polarplot with the NaN initialized data (great call!), call the figure handle's CurrentAxes property and set it to my custom axes. I then iterated and updated the line properties from polarplot and update the ThetaData and RData.

For doing the Trace-marker-along-line, I tried to create another polarplot line initialized to NaN, and then update its RData value to whatever the value was in the data, and then return the previous index value to NaN. But for some reason this was just annihilating my code and I couldn't even ctrl+c out of it. Debugger took me deep into I believe the NaN library function. Too much of a headache for today.

clf
fig = figure(1);

% Use startWin and endWin to determine where 
% to start pulling data from inputs thetaData and rData 
startWin = someStartValue;
endWin = length(thetaData);
tdat = thetaData(startWin:endWin);
rdat = rData(startWin:endWin);
timedat = time(startWin:endWin);
%

len = length(tdat);

initThetaData = NaN([1 len]); 
initRData = NaN([1 len]);
% For some reason initMarkerData = NaN([1 len]); 
% was breaking here

initThetaData(1) = tdat(1);
initRData(1) = rdat(1);

pol = polarplot(initThetaData, initRData); 
fig.CurrentAxes.RLim = [0,6];
fig.CurrentAxes.ThetaDir = 'clockwise';

startIndex = 2;
for i = startIndex:len 
    pol.ThetaData(i) = tdat(i); 
    pol.RData(i) = rdat(i);
    drawnow
    timedat(i)
end

[dpb]

As said before, polar coordinates are a sorry mess in Matlab; they simply have not developed a fully functional capability; everything is a patchwork interface to the underlying cartesian HG2 coordinate system. Apparently whoever designed/built HG/HG2 didn't do anything internally in it to handle polar/spherical coordinates. I don't know if that was developed internally by TMW or whether they bought it from somebody.

That aside, to just move a point, can't you simply just update RData, ThetaData (what a pit(proverbial)a(ppendage) that is to have to write every time! :( ) each iteration without trying to reset the earlier point(s)? The only reason you would need NaN would be if this new line was to also have a value every point the same size as the other lines.

Now, where I don't know what happens in the polar object is about whether they've built in some presumptions about multiple lines.

[Mark Duarte]

Ugh yes thank you!!! That did the trick. Movie making time.

[dpb]

Glad you got something working...the animatedline update to work with alternate functions is worth an enhancement request -- I'll use this thread as background motivation to submit it.