how to set “camera position” for 3d plots using py

问题:

I'm learning how to use mplot3d to produce nice plots of 3d data and I'm pretty happy so far. What I am trying to do at the moment is a little animation of a rotating surface. For that purpose, I need to set a camera position for the 3D projection. I guess this must be possible since a surface can be rotated using the mouse when using matplotlib interactively. But how can I do this from a script? I found a lot of transforms in mpl_toolkits.mplot3d.proj3d but I could not find out how to use these for my purpose and I didn't find any example for what I'm trying to do.

回答1:

By "camera position," it sounds like you want to adjust the elevation and the azimuth angle that you use to view the 3D plot. I've used the below script to first create the plot, then I determined a good elevation, or elev, from which to view my plot. I then adjusted the azimuth angle, or azim, to vary the full 360deg around my plot, saving the figure at each instance (and noting which azimuth angle as I saved the plot). For a more complicated camera pan, you can adjust both the elevation and angle to achieve the desired effect.

    from mpl_toolkits.mplot3d import Axes3D
    ax = Axes3D(fig)
    ax.scatter(xx,yy,zz, marker='o', s=20, c="goldenrod", alpha=0.6)
    for ii in xrange(0,360,1):
        ax.view_init(elev=10., azim=ii)
        savefig("movie%d.png" % ii)

回答2:

What would be handy would be to apply the Camera position to a new plot. So I plot, then move the plot around with the mouse changing the distance. Then try to replicate the view including the distance on another plot. I find that axx.ax.get_axes() gets me an object with the old .azim and .elev.

IN PYTHON...

axx=ax1.get_axes()
azm=axx.azim
ele=axx.elev
dst=axx.dist       # ALWAYS GIVES 10
#dst=ax1.axes.dist # ALWAYS GIVES 10
#dst=ax1.dist      # ALWAYS GIVES 10

Later 3d graph...

ax2.view_init(elev=ele, azim=azm) #Works!
ax2.dist=dst                       # works but always 10 from axx

EDIT 1... OK, Camera position is the wrong way of thinking concerning the .dist value. It rides on top of everything as a kind of hackey scalar multiplier for the whole graph.

This works for the magnification/zoom of the view:

xlm=ax1.get_xlim3d() #These are two tupples
ylm=ax1.get_ylim3d() #we use them in the next
zlm=ax1.get_zlim3d() #graph to reproduce the magnification from mousing
axx=ax1.get_axes()
azm=axx.azim
ele=axx.elev

Later Graph...

ax2.view_init(elev=ele, azim=azm) #Reproduce view
ax2.set_xlim3d(xlm[0],xlm[1])     #Reproduce magnification
ax2.set_ylim3d(ylm[0],ylm[1])     #...
ax2.set_zlim3d(zlm[0],zlm[1])     #...