CODEWITHSUNDEEP
3dmeshmeshlabsimplification
anamss
anamss

Reputation: 23

decimate 3D mesh with colors

I'm using Meshlab software to decimate 3D meshes. This works fine, however when I want to decimate a 3D mesh with colors on vertices, I don't know which algorithm to use since no one is presented managing colors (MC Edge collapse, Clustering Decimation, Quadric Edge Collapse Decimation). Any advice for decimating 3D mesh with colors? Also, I will be interested if you know some code doing that. Thanks

Upvotes: 0

Views: 559

Answers (3)

mmusy
mmusy

Reputation: 1337

Another example similar to Huang's one could be:

from vedo import Sphere

# a test mesh
s = Sphere().lw(1)

# points as numpy array
pts = s.points()
# associate a scalar to each mesh vertex
# and set its color by a standard colormap
s.cmap("rainbow", pts[:,0]).print()

s.decimate(method='pro')
s.smooth() #optionally smooth the mesh

s.show(axes=1)

enter image description here

Upvotes: 0

huang ting shieh
huang ting shieh

Reputation: 36

I have the experience about this.

First, train a KNN model for origin mesh

Second, downsample origin mesh to mesh2

Third, use KNN model to predict the mesh2 label


    from sklearn.neighbors import KNeighborsClassifier
    import vedo
    import numpy as np

    mesh = vedo.load(your_3Dobject_path)
    neigh = KNeighborsClassifier(n_neighbors=3)
    neigh.fit(mesh.cellCenters(), np.ravel(mesh.getCellArray('Label')))
    
    #downsample to 1000
    target_num = 1000
    ratio = target_num/mesh.NCells() # calculate ratio
    mesh2 = mesh.clone()
    mesh2.decimate(fraction=ratio)
    
    fine_labels = neigh.predict(mesh2.cellCenters())
    fine_labels = fine_labels.reshape(-1, 1)
    mesh2.addCellArray(fine_labels, 'Label')

Upvotes: 1

GeometryHub
GeometryHub

Reputation: 21

You can compute vertex attributes like color when doing edge collapsing.

Upvotes: 0

Related Questions