Taichi is a physically based computer graphics library, with various simulation and rendering algorithms supported (What's inside?). It's written in C++14 and wrapped friendly with Python.

Linux, Mac OS	Windows	Chat

Gallery (More...)

Installation

Step 1: Install Prerequisites

On Ubuntu 16.04

sudo apt-get install python3 git build-essential cmake make g++ libtbb-dev alien dpkg-dev debhelper ffmpeg python3-tk python3-pip

Install embree and tbb:

cd build
sudo sh ../install_embree.sh

On Arch Linux

sudo pacman -S python3 python-pip make cmake intel-tbb embree ffmpeg tk

Step 2:

Append to your ~/.bashrc:

export TAICHI_ROOT_DIR=/home/yuanming/repos/                  
export PYTHONPATH=$PYTHONPATH:$TAICHI_ROOT_DIR/taichi/python

Step 3:

Start taichi: (this will automatically install required python packages and build taichi.)

$ python3

>>> import taichi as tc
>>> ...

(Research Projects should be put into the folder projects, and will be automatically detected if the folder contains a CMakeLists.txt.)

View results

python3 python/examples/server/main.py 

Then open page http://localhost:1111. Choose a simulation output, press P to play, R to change frame rate (faster/slower).

Examples

Please see examples.

Acknowledgements

Taichi, like many other open-source projects, is based on other open-source projects:

• stb_image, stb_image_write, stb_truetype
• tinyobjloader
• fmt
• spdlog
• Catch2
• JIXIE::ImplicitQRSVD
• dcraw
• Intel Embree
• Intel TBB

Note that all of them are bundled in taichi (in source (header)/binary forms), and users do not have to manually install them.