path-tracing 0.0.1

A fast path tracing renderer

Path Tracing (WIP)

A path tracing renderer in Python accelerated by OpenMP & CUDA.

100,000 spp (1.5 min)	50,000 spp (21 min)

Getting Started

First of all, clone the repo recursively to your local environment.

git clone --recursive https://github.com/li-plus/path-tracing.git && cd path-tracing

If you miss the --recursive argument, you could still fetch the submodules in this way.

git clone https://github.com/li-plus/path-tracing.git && cd path-tracing
git submodule update --init

Python API

Install the python package from source.

sudo apt install gcc g++ cmake libeigen3-dev python3-dev python3-pip
pip install .

Run the Python example.

cd examples/python
python3 main.py --device cuda --num-samples 1000 --save-path scene.png --scene cornell_box

C++ API

C++ API is not yet stable. Please consider using Python API instead for the time being.

Install dependencies needed to build the C++ example.

sudo apt install gcc g++ cmake libeigen3-dev libopencv-dev libgflags-dev

Build the path tracing library and the C++ example.

mkdir -p build && cd build
cmake .. && make -j

Run the C++ example.

./bin/pt -device cuda -num_samples 1000 -save_path scene.png -scene cornell_box

Docker

For those who do not have root privilege, we offer an alternative to build & run path tracing within a docker container. For GPU support, nvidia-docker should be further installed.

Build a docker image with everything installed for development.

docker build ./docker -t pt:latest

Start a docker container, with project directory mapped to /opt/path-tracing.

docker run -it --rm --runtime nvidia -v $PWD:/opt/path-tracing -w /opt/path-tracing pt:latest bash

References

• smallpt
• scratchapixel
• Ray Tracing in One Weekend
• Schlick's approximation
• CUDA Texture Funcions