How to avoid unintended "jaggedness" in 3D Plotly plots?

Question

I’m using Plotly to plot a 3D dataset and the best-fit line to that dataset (Plotly version 6.2.0, Mac OS Monterey). When I plot the data in its original coordinates, the dataset and best-fit line appear jagged/wavy/sinuous (see screenshots below). However, if I simply subtract the mean from the dataset and run the exact same plotting functions, the line and dataset appear exactly how I would expect.

My best guess is this has to do with the large dataset values and orders-of-magnitude differences in the x, y, and z dimensions of the dataset. From what I’ve read online, one possible way to smooth the appearance of the line would be to use a “spline” setting for the “line_shape” attribute. However, I would like the best-fit line and data points to appear collinear with their original values and would prefer to avoid interpolating to force smoothness, if possible.

I've tried making the spacing of the axes finer and have considered using SciPy's LinearNDInterpolator, but as mentioned, I am hoping to avoid artificial smoothing of the dataset if at all possible.

Screenshot of the original dataset, showing waviness: Screenshot of wavy line

Screenshot of the same dataset with the mean removed, showing the smoothness I would expect to observe: Screenshot of same dataset with mean removed and no waviness

The relevant snippet of the original code is below. I’ve included the dataset I’m using and have formatted it exactly as it appears in my original code.

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

### Part one: Original dataset ###
# Original dataset
x = [
    242841.51371370474,
    242842.2321057521,
    242842.98216472793,
    242843.7898302282,
    242844.67710247508,
    242845.65306961897,
    242846.62956138235,
    242847.6035021302,
    242848.5577261343,
    242849.4414346082,
    242850.35453927377,
    242851.28328216364,
    242851.98668249528,
    242852.68263891042,
    242853.5190912554,
    242854.48127683587,
    242855.2495724535,
    242855.8670268352,
    242856.67226444918,
    242857.19432551065,
    242858.0543204225,
    242858.91632437005,
    242859.84946477137,
    242860.8340063954,
    242861.76712610642,
    242862.47551305266,
    242863.28264827665,
    242864.24918054775,
    242865.1861166324,
    242865.89921764264,
    242865.96585560686,
    242865.54343995376,
    242866.04327942675,
    242866.95609290188,
    242867.93530624994,
    242868.87682222814,
    242869.6040531725,
    242870.2818274008,
    242871.195308515,
    242872.02078952064,
    242872.80658931847,
    242873.68504109397,
    242874.63072940378,
    242875.61506473337,
    242876.59956580945,
    242877.56710429245,
    242878.4417633086,
    242879.31683372613,
    242880.13168928024,
    242880.92023267024,
    242881.71379591143,
    242882.54724389955,
    242883.4482578614,
    242884.30376656697,
    242885.19765295103,
    242886.11825244414,
    242886.9600141796,
    242887.36266717812,
    242888.00615052023,
    242888.76926449002,
    242889.5641237527,
    242890.47554397324,
    242891.45691368319,
    242892.43368979736,
    242893.4122051768,
    242894.29546648718,
    242895.14065051088,
    242896.09998258963,
    242897.0531311911,
    242897.89569041005,
]
y = [
    3906806.867605061,
    3906806.99149825,
    3906807.120852687,
    3906807.2601418886,
    3906807.413159992,
    3906807.5814743303,
    3906807.749879144,
    3906807.9178440124,
    3906808.08240855,
    3906808.2348120487,
    3906808.392285186,
    3906808.552455276,
    3906808.6737630083,
    3906808.793786971,
    3906808.9380407236,
    3906809.103978307,
    3906809.2364778174,
    3906809.3429633956,
    3906809.4818338864,
    3906809.571868026,
    3906809.720181907,
    3906809.868842264,
    3906810.0297707445,
    3906810.1995638297,
    3906810.3604887417,
    3906810.482656461,
    3906810.6218542117,
    3906810.7885414213,
    3906810.9501245017,
    3906811.073105204,
    3906811.084597522,
    3906811.011748132,
    3906811.0979499584,
    3906811.2553728768,
    3906811.4242470525,
    3906811.5866199764,
    3906811.7120375135,
    3906811.8289257935,
    3906811.9864638527,
    3906812.1288254987,
    3906812.264343763,
    3906812.4158406965,
    3906812.5789331766,
    3906812.7486906843,
    3906812.918476776,
    3906813.0853375164,
    3906813.2361803544,
    3906813.387094142,
    3906813.5276233335,
    3906813.6636147546,
    3906813.8004718944,
    3906813.944207519,
    3906814.099595505,
    3906814.2471356993,
    3906814.40129447,
    3906814.5600601574,
    3906814.7052295627,
    3906814.7746707047,
    3906814.885645212,
    3906815.0172511004,
    3906815.154331751,
    3906815.31151439,
    3906815.48076045,
    3906815.6492143027,
    3906815.8179681073,
    3906815.9702944886,
    3906816.116054098,
    3906816.2814995693,
    3906816.4458786445,
    3906816.591185583,
]
z = [
    663.4151597804555,
    663.4437289572488,
    663.4735574715868,
    663.5056768951674,
    663.540962137783,
    663.5797746190462,
    663.6186079634939,
    663.6573398585826,
    663.6952876537634,
    663.7304311714518,
    663.7667437235136,
    663.8036781800032,
    663.8316511628669,
    663.8593281143825,
    663.89259233969,
    663.9308567526336,
    663.9614105058508,
    663.9859655708151,
    664.0179884416671,
    664.0387498835784,
    664.0729503549891,
    664.1072307221838,
    664.1443400598992,
    664.1834935330202,
    664.2206020479188,
    664.2487733395961,
    664.2808716750297,
    664.3193089481462,
    664.3565692334507,
    664.3849279950878,
    664.3875780686622,
    664.3707793477998,
    664.3906570765743,
    664.4269580485095,
    664.4658996255514,
    664.5033420450945,
    664.5322627291333,
    664.5592166073463,
    664.595544130102,
    664.6283720447158,
    664.6596219081147,
    664.6945563762371,
    664.7321647220051,
    664.7713099911615,
    664.8104618517616,
    664.8489391401395,
    664.883722776953,
    664.9185227744699,
    664.9509281337324,
    664.9822871049239,
    665.0138457066884,
    665.0469904540547,
    665.0828221802874,
    665.1168442432357,
    665.1523925183491,
    665.1890031264005,
    665.2224784967451,
    665.2384912919068,
    665.2640814824322,
    665.2944291707121,
    665.3260393129514,
    665.3622848776236,
    665.4013122093502,
    665.4401568619255,
    665.4790706819934,
    665.5141964167825,
    665.5478078854554,
    665.5859588196997,
    665.6238638480258,
    665.657370932874,
]

# Dataset converted into dictionary and Pandas DataFrame
data_dict = {"x": x, "y": y, "z": z}
test_data = pd.DataFrame(data_dict)

# Calculate mean of dataset
data_mean = [test_data["x"].mean(), test_data["y"].mean(), test_data["z"].mean()]

# Best-fit 3D line to dataset
vector_in = np.array([11.92932516, 2.05731974, 0.47440809])

# Plot best-fit line
linearspacing = np.linspace(-100, 100, 1000)
plot_line_x = data_mean[0] + (vector_in[0] * linearspacing)
plot_line_y = data_mean[1] + (vector_in[1] * linearspacing)
plot_line_z = data_mean[2] + (vector_in[2] * linearspacing)

# Plot of dataset and line - line and dataset appear wavy/jagged
test_figure = go.Figure(
    data=[
        go.Scatter3d(
            x=test_data["x"],
            y=test_data["y"],
            z=test_data["z"],
            name="Data",
            mode="markers",
            marker=dict(
                size=5, line=dict(width=2, color="Black"), color="cyan", opacity=0.5
            ),
        )
    ]
)
test_figure.add_trace(
    go.Scatter3d(
        x=plot_line_x,
        y=plot_line_y,
        z=plot_line_z,
        mode="lines",
        name="Best-fit Line",
        line=dict(color="black", width=3),
    )
)
test_figure.update_layout(
    title="Original data; best-fit line and datapoints appear wavy",
    scene=dict(
        xaxis=dict(range=[data_mean[0] - 40, data_mean[0] + 40]),
        yaxis=dict(range=[data_mean[1] - 40, data_mean[1] + 40]),
        zaxis=dict(range=[data_mean[2] - 40, data_mean[2] + 40]),
    ),
)
test_figure.show()

### Part two: Original dataset with mean subtracted ###
# Same dataset with mean subtracted
x2 = x - data_mean[0]
y2 = y - data_mean[1]
z2 = z - data_mean[2]

# Mean-subtracted dataset converted into dictionary and Pandas DataFrame
data_dict2 = {"x2": x2, "y2": y2, "z2": z2}
test_data2 = pd.DataFrame(data_dict2)

# Same vector plotted, but centered at origin
plot_line_x2 = vector_in[0] * linearspacing
plot_line_y2 = vector_in[1] * linearspacing
plot_line_z2 = vector_in[2] * linearspacing

# Plot of dataset and line
test_figure2 = go.Figure(
    data=[
        go.Scatter3d(
            x=test_data2["x2"],
            y=test_data2["y2"],
            z=test_data2["z2"],
            name="Data",
            mode="markers",
            marker=dict(
                size=5, line=dict(width=2, color="Black"), color="cyan", opacity=0.5
            ),
        )
    ]
)
test_figure2.add_trace(
    go.Scatter3d(
        x=plot_line_x2,
        y=plot_line_y2,
        z=plot_line_z2,
        mode="lines",
        name="Best-fit Line",
        line=dict(color="black", width=3),
    )
)
test_figure2.update_layout(
    title="Data with mean subtracted; no waviness",
    scene=dict(
        xaxis=dict(range=[-40, 40]),
        yaxis=dict(range=[-40, 40]),
        zaxis=dict(range=[-40, 40]),
    ),
)
test_figure2.show()

I've posted the same question to Plotly's community forum as well (and will update with any answers that solve the issue).

Your z-axis range is inappropriate for the data (in both plots). For the "wavy" one change it to zaxis=dict(range=[data_mean[2] - 1, data_mean[2] + 1]) — lastchance
– lastchance, Commented Aug 7 at 9:03

Stitt · Accepted Answer · 2025-08-07 07:35:04Z

After some investigation, I believe you're seeing issues due to either floating point or rounding error on the plot, possibly due to the significant order-of-magnitude differences between the values on each axis.

I got around this by changing your plot_line_(x/y/z) lists from being 1000 elements to just 2 (the first and last). That way, Plotly just has to draw a single line between two points, rather than 1000 lines which will introduce the errors you're seeing. The only other change I made to facilitate this was reducing the line_width of the line to 0.01 (I'm unsure exactly what scale the line width is measured against, and if it is perhaps relative to the length of line). Leaving the value at 3 created an extremely thick line.

Summary of changes:

plot_line_x = data_mean[0] + (vector_in[0] * (-100, 100))
plot_line_y = data_mean[1] + (vector_in[1] * (-100, 100))
plot_line_z = data_mean[2] + (vector_in[2] * (-100, 100))

Collectives™ on Stack Overflow

How to avoid unintended "jaggedness" in 3D Plotly plots?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related