3D Visualizations¶

ggplotly provides geoms for creating interactive 3D plots powered by Plotly's WebGL renderer.

3D Scatter Plots¶

Basic 3D Scatter¶

In [1]:

import numpy as np
import pandas as pd
from ggplotly import *

df = pd.DataFrame({
    'x': np.random.randn(200),
    'y': np.random.randn(200),
    'z': np.random.randn(200)
})

(ggplot(df, aes(x='x', y='y', z='z')) + geom_point_3d())

import numpy as np import pandas as pd from ggplotly import * df = pd.DataFrame({ 'x': np.random.randn(200), 'y': np.random.randn(200), 'z': np.random.randn(200) }) (ggplot(df, aes(x='x', y='y', z='z')) + geom_point_3d())

Out[1]:

Colored by Group¶

In [2]:

df = pd.DataFrame({
    'x': np.random.randn(200),
    'y': np.random.randn(200),
    'z': np.random.randn(200),
    'group': np.random.choice(['A', 'B', 'C'], 200)
})

(ggplot(df, aes(x='x', y='y', z='z', color='group')) + geom_point_3d(size=6))

df = pd.DataFrame({ 'x': np.random.randn(200), 'y': np.random.randn(200), 'z': np.random.randn(200), 'group': np.random.choice(['A', 'B', 'C'], 200) }) (ggplot(df, aes(x='x', y='y', z='z', color='group')) + geom_point_3d(size=6))

Out[2]:

Colored by Continuous Variable¶

In [3]:

df = pd.DataFrame({
    'x': np.random.randn(200),
    'y': np.random.randn(200),
    'z': np.random.randn(200),
    'value': np.random.rand(200) * 100
})

(ggplot(df, aes(x='x', y='y', z='z', color='value'))
 + geom_point_3d(size=5)
 + scale_color_gradient(low='blue', high='red'))

df = pd.DataFrame({ 'x': np.random.randn(200), 'y': np.random.randn(200), 'z': np.random.randn(200), 'value': np.random.rand(200) * 100 }) (ggplot(df, aes(x='x', y='y', z='z', color='value')) + geom_point_3d(size=5) + scale_color_gradient(low='blue', high='red'))

Out[3]:

3D Surfaces¶

Creating Surface Data¶

Surfaces require gridded data. Here's a helper function:

In [4]:

def make_surface(func, x_range=(-5, 5), y_range=(-5, 5), resolution=50):
    """Generate surface data from a function z = f(x, y)."""
    x = np.linspace(x_range[0], x_range[1], resolution)
    y = np.linspace(y_range[0], y_range[1], resolution)
    X, Y = np.meshgrid(x, y)
    Z = func(X, Y)
    return pd.DataFrame({
        'x': X.flatten(),
        'y': Y.flatten(),
        'z': Z.flatten()
    })

def make_surface(func, x_range=(-5, 5), y_range=(-5, 5), resolution=50): """Generate surface data from a function z = f(x, y).""" x = np.linspace(x_range[0], x_range[1], resolution) y = np.linspace(y_range[0], y_range[1], resolution) X, Y = np.meshgrid(x, y) Z = func(X, Y) return pd.DataFrame({ 'x': X.flatten(), 'y': Y.flatten(), 'z': Z.flatten() })

Paraboloid¶

In [5]:

df = make_surface(lambda x, y: x**2 + y**2)
(ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Viridis'))

df = make_surface(lambda x, y: x**2 + y**2) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Viridis'))

Out[5]:

Saddle Surface (Hyperbolic Paraboloid)¶

In [6]:

df = make_surface(lambda x, y: x**2 - y**2)

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='RdBu')
 + labs(title='Saddle Surface'))

df = make_surface(lambda x, y: x**2 - y**2) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='RdBu') + labs(title='Saddle Surface'))

Out[6]:

Sinc Function (2D)¶

In [7]:

def sinc_2d(x, y):
    r = np.sqrt(x**2 + y**2)
    return np.where(r == 0, 1, np.sin(r) / r)

df = make_surface(sinc_2d, x_range=(-10, 10), y_range=(-10, 10), resolution=80)

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='Plasma')
 + labs(title='2D Sinc Function'))

def sinc_2d(x, y): r = np.sqrt(x**2 + y**2) return np.where(r == 0, 1, np.sin(r) / r) df = make_surface(sinc_2d, x_range=(-10, 10), y_range=(-10, 10), resolution=80) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Plasma') + labs(title='2D Sinc Function'))

Out[7]:

Trigonometric Surface¶

In [8]:

df = make_surface(lambda x, y: np.sin(x) * np.cos(y))

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='Viridis')
 + labs(title='sin(x) * cos(y)'))

df = make_surface(lambda x, y: np.sin(x) * np.cos(y)) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Viridis') + labs(title='sin(x) * cos(y)'))

Out[8]:

Surface Colorscales¶

Available colorscales for geom_surface:

• Sequential: Viridis, Plasma, Inferno, Magma, Cividis, Blues, Greens, Reds, YlOrRd, YlGnBu
• Diverging: RdBu, RdYlBu, RdYlGn, BrBG, PiYG, PRGn, Spectral
• Other: Jet, Hot, Electric, Blackbody, Earth, Picnic, Portland

Wireframe Plots¶

Wireframes show the surface structure without solid fills:

In [9]:

df = make_surface(lambda x, y: np.sin(x) * np.cos(y), resolution=30)

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_wireframe(color='steelblue', linewidth=1)
 + labs(title='Wireframe Plot'))

df = make_surface(lambda x, y: np.sin(x) * np.cos(y), resolution=30) (ggplot(df, aes(x='x', y='y', z='z')) + geom_wireframe(color='steelblue', linewidth=1) + labs(title='Wireframe Plot'))

Out[9]:

Wireframe Parameters¶

Parameter	Default	Description
color	'steelblue'	Line color
linewidth	1	Line width
opacity	1.0	Transparency (0-1)

Combining 3D Geoms¶

You can layer 3D geoms:

In [10]:

# Surface with scatter points
df_surface = make_surface(lambda x, y: np.sin(x) * np.cos(y))

# Sample points on the surface
sample_idx = np.random.choice(len(df_surface), 50, replace=False)
df_points = df_surface.iloc[sample_idx].copy()
df_points['z'] = df_points['z'] + 0.1  # Offset slightly above surface

(ggplot(df_surface, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='Viridis', opacity=0.7)
 + geom_point_3d(data=df_points, color='red', size=5))

# Surface with scatter points df_surface = make_surface(lambda x, y: np.sin(x) * np.cos(y)) # Sample points on the surface sample_idx = np.random.choice(len(df_surface), 50, replace=False) df_points = df_surface.iloc[sample_idx].copy() df_points['z'] = df_points['z'] + 0.1 # Offset slightly above surface (ggplot(df_surface, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Viridis', opacity=0.7) + geom_point_3d(data=df_points, color='red', size=5))

Out[10]:

Mathematical Visualizations¶

Gaussian (Bell Curve) in 3D¶

In [11]:

def gaussian_2d(x, y, sigma=1):
    return np.exp(-(x**2 + y**2) / (2 * sigma**2))

df = make_surface(gaussian_2d, x_range=(-3, 3), y_range=(-3, 3), resolution=60)

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='Viridis')
 + labs(title='2D Gaussian Distribution'))

def gaussian_2d(x, y, sigma=1): return np.exp(-(x**2 + y**2) / (2 * sigma**2)) df = make_surface(gaussian_2d, x_range=(-3, 3), y_range=(-3, 3), resolution=60) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Viridis') + labs(title='2D Gaussian Distribution'))

Out[11]:

Ripple Effect¶

In [12]:

def ripple(x, y):
    r = np.sqrt(x**2 + y**2)
    return np.sin(3 * r) * np.exp(-0.3 * r)

df = make_surface(ripple, x_range=(-5, 5), y_range=(-5, 5), resolution=80)

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='RdBu')
 + labs(title='Ripple Effect'))

def ripple(x, y): r = np.sqrt(x**2 + y**2) return np.sin(3 * r) * np.exp(-0.3 * r) df = make_surface(ripple, x_range=(-5, 5), y_range=(-5, 5), resolution=80) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='RdBu') + labs(title='Ripple Effect'))

Out[12]:

Rosenbrock Function (Optimization Test)¶

In [13]:

def rosenbrock(x, y, a=1, b=100):
    return (a - x)**2 + b * (y - x**2)**2

df = make_surface(rosenbrock, x_range=(-2, 2), y_range=(-1, 3), resolution=60)

(ggplot(df, aes(x='x', y='y', z='z'))
 + geom_surface(colorscale='Hot')
 + labs(title='Rosenbrock Function'))

def rosenbrock(x, y, a=1, b=100): return (a - x)**2 + b * (y - x**2)**2 df = make_surface(rosenbrock, x_range=(-2, 2), y_range=(-1, 3), resolution=60) (ggplot(df, aes(x='x', y='y', z='z')) + geom_surface(colorscale='Hot') + labs(title='Rosenbrock Function'))

Out[13]:

Interactivity¶

All 3D plots support:

• Rotation: Click and drag to rotate
• Zoom: Scroll wheel or pinch
• Pan: Shift + drag
• Reset: Double-click

The 3D camera position is automatically saved when you interact, so subsequent renders maintain your viewpoint.