sci3d: Lightweight Interactive Scientific Visualization library

Motivation

This library was created to cover a set of needs not fully supplied by currently available plotting libraries. It was designed with the following goals in mind:

Real-time, dynamic, easy to interact 3D plots

Matplotlib has some great 3D capabilities, but it's not very efficient for displaying animated plots. Furthermore, it's difficult to manually interact with the plots in such cases.

Plotly is another advanced plotting library with support for 3D charts. Since it renders the the plots in HTML, that makes it a great fit for working with Jupyter notebooks. However, that means plots cannot be updated dynamically after they are created, which can make it less suitable for certain interactive applications.

Easy to install, succint and self documented

Enthought's Mayavi is a very complete tool, with a wide set of plotting capabilities that can be coupled with an MVC-like library of UI widgets. However, its dependencies make it difficult to install on environments that relies on too many libraries, such as Qt and OpenCV. Additionally, it may also require a lot of boilerplate code to achieve certain goals.

sci3d seeks to avoid these issues by adopting the following principles:

• Lean set of dependencies, minimizing potential for package conflicts
• API functions are always explicit and use type hints. This way, you can infer what a method expects without leaving your IDE. Likewise, you know what can be done with objects by relying on autocomplete features.

Installation

WSL2 requirements

If running from WSL2, you will need to install the latest mesa package on your Linux system. Check this discussion for some hints on how to do that.

After installing the latest mesa, make sure you have all the requirements.

Requirements

sci3d relies on packages that have the following dependencies:

• libxrandr
• libxinerama
• libxcursor
• libxi
• libxext

On Ubuntu, these can be installed via:

sudo apt install libxrandr-dev libxinerama-dev libxcursor-dev libxi-dev libxext-dev

Finally, sci3d can be installed via pip.

Install via pip

pip install git+https://github.com/csantosbh/sci3d

Usage

Example

import numpy as np
import sci3d as s3d
import time

# Create sphere SDF
def make_sphere(radius):
    sphere = np.mgrid[-1:1:32j, -1:1:32j, -1:1:32j].astype(np.float32)
    return np.linalg.norm(sphere, 2, 0) - radius

# Plot SDF
fig = s3d.isosurface(make_sphere(0.5))

# Wait while window is open
dt = 1.0/30
t = 0
while s3d.get_window_count() > 0:
    # Animate sphere by changing its size
    fig.set_isosurface(make_sphere(abs(0.5 + 0.3 * np.cos(t))))

    time.sleep(dt)
    t += dt

# Shutdown library
s3d.shutdown()

For more advanced examples, check the demos.

Controls

• Left mouse button: Rotate camera
• Scroll: Zoom
• wasd: Translate camera fps style. Press shift to move faster.
• f: Recenter scene with respect to camera