Awesome

PhantomGaze: A GPU-Accelerated Rendering Engine for Scientific Visualization

Introduction

PhantomGaze is a Python package that provides a basic rendering engine optimized for scientific computing visualizations. Utilizing the power of Numba, it enables efficient operations directly on the GPU, enhancing the visualization of complex scientific data. This package is particularly useful for rendering volumetric data, geometric shapes, and contour visualizations with ease.

Features

GPU-accelerated rendering engine
Volumetric rendering
Geometric rendering
Contour rendering
Customizable color maps
Integration with GPU libraries such as CuPy, JAX, PyTorch, and Warp.

Installation

PhantomGaze can be installed using pip:

pip install .

(TODO: Make package available on PyPI)

Usage

There are several examples in the examples directory. A minimal example of a contour plot is shown below:

import cupy as cp
import matplotlib.pyplot as plt

import phantomgaze as pg

# Create SDF feild of a sphere using cupy
lin = cp.linspace(-1, 1, 256)
X, Y, Z = cp.meshgrid(lin, lin, lin, indexing="ij")
sphere = -(cp.sqrt(X**2 + Y**2 + Z**2) - 1.0)
sphere_volume = pg.objects.Volume(
    sphere, spacing=(2 / 256, 2 / 256, 2 / 256), origin=(-1.0, -1.0, -1.0)
)
color_volume = pg.objects.Volume(
    cp.sin(X * 2 * cp.pi) * cp.sin(Y * 2 * cp.pi) * cp.sin(Z * 2 * cp.pi),
    spacing=(2 / 256, 2 / 256, 2 / 256),
    origin=(-1.0, -1.0, -1.0),
)

# Create camera object
camera = pg.Camera(position=(2.0, 1.0, -4.0), focal_point=(0.0, 0.0, 0.0), view_up=(0.0, 1.0, 0.0))

# Render the contour of the inner sphere
screen_buffer = pg.render.contour(sphere_volume, camera, threshold=0.0, color=color_volume)

# Show the rendered image
plt.imshow(screen_buffer.image.get())
plt.show()

This produces the following image:

Contour

Gallery

The following images were generated using the examples in the examples directory.

Axes Geometry Volume