Ray Splitting for DVR

Code and datasets for "Analytic Ray Splitting for Controlled Precision DVR", Sebastian Weiss and Rüdiger Westermann, EuroVis 2021.

Currently supported:

Iso-surface rendering
- Trilinear, analytic + numerical
- Tricubic, numerical
- Multiple semi-transparent isosurfaces
Direct Volume Rendering
- trilinear, piecewise linear transfer function
- rectangle rule, trapezoid rule, Simpson rule. With fixed number of control points, variable number of control points (analytic bounds), iterative refinement (Simpson only)
- scale-invariant DVR
Hybrid of DVR+Iso

Project structure:

renderer: the library that contains the rendering core (C++, CUDA)
gui: interactive gui, allows to test all available options (C++, OpenGL)
bindings: python bindings to the renderer core (C++, Python)
tests: test cases and evaluation written in C++python. The code that generated the data for the plots in the paper

Data

volumes: all test volumes used in the plots and images. Note that the synthetic datasets (Marschner Lobb, Sphere, Tube) are are generated in the program, not loaded from external files
scenes: scene settings to replicate the images in the paper

See the releases for compiled binaries.

Requirements

CUDA
OpenGL with GLFW and GLM

Tested with CUDA 11.0 (device architecture 7.5), Windows 10, Visual Studio 2019

Video