Interpolation in plotly.graph_objects.Surface() plots


I’m plotting slices of a 3d volume with plotly.graph_objects.Surface(). As a standard the surface appears to undergo some kind of interpolation when plotted. See figure below.

I cannot see that it is described anywhere in the documentation, which makes me doubt what is actually going on.

But how do I get rid of this “interpolation”? I wish to show my true data.

Best regards.


Why do you suspect that the slice points are incorrectly colormapped?

Let us say that you get the slice sectioning the volume through the plane x=2. Within the resulting planar shape
you have induced a meshgrid from your volume. If at at given (y,z) in the slice the associated value is val, then exactly this val is mapped to a color in the colorscale. But since your planar shape has a discrete representation (consisting in the points of intersection of the discrete volume with x=2), at the points that are not in the induced meshgrid, the color is interpolated.


I thought I understood your point. If the grid of the meshgrid is not aligned with the grid of the volume (array), then interpolation will occur?

But according to the example below, I do not understand your point.

import numpy as np
import plotly 
import plotly.graph_objects as go

n = 100
l = 10.0
r = 3.0

# create empty volume
volume = np.zeros((n, n, n))

# draw a cylinder within the volume
x = y = np.linspace(-l/2, l/2, n)
X, Y = np.meshgrid(x, y)

volume[X**2 + Y**2 < r] = 1

# plot
Z = np.zeros(np.shape(volume)[:2])

cross_section = go.Surface(z = Z,
                           x = list(X),
                           y = list(Y),
                           surfacecolor=volume[:, :, 5],)

fig = go.Figure(data=cross_section)
plotly.offline.plot(fig, filename='test.html')

In this case the meshgrid is aligned with the grid that corresponds to the volume/array. The volume contains only zeros and ones. I still see interpolated values in the plot.

Best regards.


Let me explain how the colormapping of some data to a Plotly colorscale works.

First the colorscale is defined by a scale, i.e. a list of ordered values in [0,1]:
and a list of color codes, one for each scale entry:

my_colorscale = [[0.0, 'rgb(253, 253, 204)'],
                 [0.1, 'rgb(201, 235, 177)'],
                 [0.2, 'rgb(145, 216, 163)'],
                 [0.3, 'rgb(102, 194, 163)'],
                 [0.4, 'rgb(81, 168, 162)'],
                 [0.5, 'rgb(72, 141, 157)'],
                 [0.6, 'rgb(64, 117, 152)'],
                 [0.7, 'rgb(61, 90, 146)'],
                 [0.8, 'rgb(65, 64, 123)'],
                 [0.9, 'rgb(55, 44, 80)'],
                 [1.0, 'rgb(39, 26, 44)']]

The scale is the list of values [0, 0.1, 0.2, .....1].
Now suppose that your trace defintion colormaps uni-dimensional data, vals , of range [valmin, valmax] .
First plotly.js normalizes vals, i.e. maps them via the normalization function:

val --> norm_val=(val-valmin)/(valmax-valmin) in [0,1]

if the norm_val coincides with one of the scale entry, then to val is assigned the corresponding color in my_colorscale. Otherwise, plotly.js finds the interval of consecutive values in the scale, that contains the norm_val and calculates the corresponding color by interpolating linearly the color codes corresponding the the interval ends.

From your posted image and the red arrow I do not understand how would you like to get the colormapping.

Could you point out a surface image on the web plotted with another tool that maps values to colors and no interpolation is performed?