vtkScalarBarActor usage

Support
1

Hello Everyone,

I’m using vedo to create a network with nodes and edges and color the nodes using the scalar values stored in an array. From what I understand, vedo uses vtkScalarBarActor to associate the scalar values to colors.
Code:

import networkx as nx
from vedo import *

G = nx.gnm_random_graph(n=10, m=15, seed=1)
nxpos = nx.spring_layout(G)
nxpts = [nxpos[pt] for pt in sorted(nxpos)]

nx_lines = []
for i, j in G.edges():
    p1 = nxpos[i].tolist() + [0]  # add z-coord
    p2 = nxpos[j].tolist() + [0]
    nx_lines.append([p1, p2])

nx_pts = Points(nxpts, r=12)
nx_edg = Lines(nx_lines).lw(2)

# node values
vals = [100, .80, .10, .79, .70, .60, .75, .78, .65, .90]
nx_pts.pointColors(vals, cmap='YlGn', vmin=min(vals), vmax=max(vals)).addScalarBar()
show(nx_pts, nx_edg, nx_pts.labels('id'), interactive=True, bg='black', title='plot')

I’m facing a problem when there are outliers in the scalar values.

image
image719×720 40.2 KB

vals = [100, .80, .10, .79, .70, .60, .75, .78, .65, .90] in here there is one outlier and due to this the rest of the values are assigned a single color. I would like to ask for suggestions on how to avoid the above and create a scalarbar like the following
image

The red discrete patch in the bottom representation is for the outliers (i.e 100 in vals = [100, .80, .10, .79, .70, .60, .75, .78, .65, .90])

More details on the same issue can be found here

2

Hello,

I suggest you to study or find examples using vtkColorTransferFunction class: https://vtk.org/doc/nightly/html/classvtkColorTransferFunction.html. You programmaticaly build it with calls to its addRGBSegment() method. First, make sure the transfer function is clear by making a prior call to RemoveAllPoints(). For the valid continous values you do something like addRGBSegment(0.7, r1, g1, b1, 0.8, r2, g2, b2). For the outlier code (e.g. 100) you do like addRGBSegment(99.5, r, g, b, 100.5, r, g, b). Notice the usage of the same RGB values for the range that includes the 100 value.

After your transfer function is defined, you build a LUT with this recipe (in C++, sorry):

    //create the color table object
    vtkSmartPointer<vtkLookupTable> lut = vtkSmartPointer<vtkLookupTable>::New();
    lut->SetTableRange(0.0, 100.5);
    lut->SetNumberOfTableValues(64);
    for(size_t i = 0; i < tableSize; ++i)
    {
        double *rgb;
        double value = /* lookup/compute value from i and 0.0-1.0; 100 values */;
        rgb = transferFunction->GetColor( value ); //value can be 0.0-1.0 and 100
        lut->SetTableValue(i, rgb[0], rgb[1], rgb[2]);
    }
    lut->SetRampToLinear(); //you may play with other interpolation methods
    lut->Build();

I hope this helps.

regards,

Paulo

1 Like
3

Hi Paulo, Thank you so much for the details.

@marcomusy Could you please check if this will be helpful?