Returning variables from FreeHandCutPlotter parent class

[jkissi]

Hi there Marco!!
So I have finally made it onto that selection window question I asked, and am trying to get my head around the workings of the cut plotter. I decided to make a new thread about it, I think thats a good idea.
I have decided to extend the FreeHandCutPlotter class, rather than write something from scratch because similar functionality to that which I need is already in there (once again, thanks for an awesome job you've done with this library, its so good!!) However, I am running into problems getting the values from the super class back for examination and was hoping you might have some insight as to what I am doing wrong here.

I need to understand exactly how the class segments the points based on the user interaction, so that I can extend it by adding methods for cutters of different geometries. This is what I have so far:

class newCutPlotter(vedo.applications.FreeHandCutPlotter):
     """Test for the creation of a class extension for cutting point clouds
     """

     def __init__(self, mesh, **options):
         super().__init__(mesh, **options)


     def newFunction(self, *args, **kwargs):
         """Function that is new!!!
         """
         super().start(*args, **kwargs)

         print("Skip all the way home")


     def getPointsCallback(self, evt):
         """Call back to get the points under the cursor
         """
         if not evt.actor:
             return
         print(f"These are the cpoints: {self.cpoints}")



     def getSplineValsCallback(self, evt):
         """Call back to get the points in the spline
         """

         tol = self.mesh.diagonalSize()/2            # size of ribbon (not shown)
         pts = self.spline.points()
         n = fitPlane(pts, signed=True).normal       # compute normal vector to points
         rb = Ribbon(pts - tol*n, pts + tol*n, closed=True)
         #self.mesh.cutWithMesh(rb, invert=inv)       # CUT

         print(f"These are the tolerance points: {tol}")
         print(f"These are the spline points: {pts}")



 vedo.settings.useParallelProjection = True  # to avoid perspective artifacts

 pts = np.random.randn(10000, 3)
 pts = Points(pts).ps(8)
 pts.name = "My Gaussian"

 cube = Cube().x(2)

 plt = newCutPlotter(pts).addHoverLegend()
 plt.addCallback('RightButtonRelease', plt.getPointsCallback)
 plt.addCallback('KeyPress', plt.getSplineValsCallback)
 plt.newFunction(pts, cube, bg2='lightblue',axes=1)


 print('Done')
 print('Done')

This code produces the following error:

  File "D:\installs\programs\Anaconda\Anaconda3\envs\nnse\lib\site-packages\vedo\plotter.py", line 1708, in _func_wrap
    func(event_dict)

  File "d:\life\canada\uwo\ece\phd\research\code\nnse\testo3d.py", line 1168, in getSplineValsCallback
    pts = self.spline.points()

AttributeError: 'NoneType' object has no attribute 'points'

I am confused by this. How can self.spline be None, while self.mesh has a value? If I remove getSplineValsCallback(), then the cutter runs with full functionality, so I don't think there is a problem with my extension of the class. Maybe you have some insights as to what I am doing wrong here?

Cheers
j

[marcomusy]

Hi @jkissi
try
plt.addCallback('KeyPress', plt.getSplineValsCallback, priority=100)
so the object is not erased by the previous z keypress call of the parent.
Also
super().start(*args, **kwargs)
looks a bit weird to my eye.. should be better
self.start(*args, **kwargs)

[marcomusy]

Am I right in thinking that the self.spline points are the points that should remain after the cutting action?

uhm no, those are the points of the spline (upsampled wrt cpoints if i remember)..
the pts Mesh/Points object is cut in-place

And also that self.cpoints are the coordinates of the points that make up the cutting line?

Yes

Btw note that you can add a plt.start()at the end of the script to repeat the cutting.

[jkissi]

so are you saying that the coordinates of the points that are to be removed (or remain) are not stored in any variables in this class? I am asking because that is something I will need to do for my extension....

[marcomusy]

I'm afraid not! It's just cut and gone... you would like to mark the points instead of removing them?

Option 1
You may rewrite just the keypress call :
plt.removeCallback(plt.idkeypress)

and write a custom version where instead of cutWithMesh() you use insidePoints() and then .threshold()
analogous to this

Option 2
I can easily change those lines but you'll need to wait for the next release.

[jkissi]

Thanks for the offer Marco but I cannot wait for the next release, so I will attempt to extend it with your suggestion. I will fiddle for a bit and get back to you...hope you're having a great day!

j

[jkissi]

Actually, let me rephrase that: I think it's a great idea to do that for the next release Marco but I will also be looking at your first suggestion in the mean time ;-)

[jkissi]

Hi there Marco,
I have been messing around with your suggestions and am having some success. I just wanted to ask: precisely why I am using the threshold() function? Using the insidepoints() function on the ribbon object gives the points....what added functionality does threshold() provide?

[marcomusy]

Ouch I'm afraid I've found a small bug in the insidepoints() function..
And - you are right - you dont' really need the threshold()

After
pip install -U git+https://github.com/marcomusy/vedo.git

you should be able to reproduce this:

from vedo import *

settings.useDepthPeeling = True

pts = Points(np.random.randn(10000, 3)*[3,2,1] + [50,60,70])

elli = pcaEllipsoid(pts, pvalue=0.50)

elli.insidePoints(pts, returnIds=True)

ids  = elli.insidePoints(pts, returnIds=True)
pts.print()  # a new "IsInside" array now exists in pts
pin = pts.points()[ids]
print("inside  points #", len(pin))

# Create an inverted mask instead of calling insidePoints(invert=True)
mask = np.ones(pts.N(), dtype=bool)
mask[ids] = False
pout = pts.points()[mask]
print("outside  points #", len(pout))

show(elli,
     Points(pin).c("green4"),
     Points(pout).c("red5").alpha(0.2),
     axes=1,
).close()

[jkissi]

great, thats awesome!!