Open source
Taichi provides open source implementation of state-of-the-art computer graphics research results.
State-of-the-art
With various latest algorithms included in Taichi, researchers and artists can enjoy handy comparison with and easy utilization of these new technologies.
Easy-to-use
Though the kernel part is implemented in C++11 for efficiency, Python bindings are provided for fast scripting and content generation.
What's Inside
Currently I implemented about 40 graphics papers, most of which are from SIGGRAPH(Asia)/ToG.
Global Illumination Algorithms Supported:
- (Volumetric) Path Tracing ((V)PT)
- Light Tracing (LT)
- Photon Mapping (PM)
- (Stochastic/Probabilistic) Progressive Photon Mapping (SPPM/PPPM)
- Unified Path Sampling (UPS)/Vertex Connection and Merging (VCM)
- Bidirectional Path Tracing (BDPT)
- Bidirectional Photon Mapping (BDPM)
- Primary Sample Space Metropolis Light Transport (PSSMLT) with BDPT
- PSSMLT with Volumetric PT
- Multiplexed Metropolis Light Transport (MMLT)
- Adaptive Markov Chain Monte Carlo Progressive Photon Mapping (AMCMCPPM)
- Markov Chain Monte Carlo UPS/VCM
Physically based Animation Algorithms Supported:
- Eulerian Fluids (or Stable Fluids)
- Particle-in-Cell Fluids (PIC)
- Fluid Implicit Particles (FLIP)
- Affine Particle-in-Cell (APIC)
- Material Point Method for snow (MPM)
- MPM for sand
- Weakly Compressible Smoothed Particle Hydrodynamics (WCSPH)
- Predictive-Corrective Incompressible SPH (PCISPH)
- (Modified) Incomplete Cholesky Preconditioned Conjugate Gradient ((M)ICPCG) Poisson Solver
- Multigrid PCG Poisson Solver
- Impulse-based Rigid Body Simulation (real time in 2D)
Other Features:
- Multithreading
- Seamless data transfer from simulation to rendering
- Scripting interface for procedural content generation
- ...