Monte Carlo Path Tracer

Features

• Monte Carlo path tracing. (path_tracing.cpp)
  • With Russian Roulette method.
• Multiple importance sampling. (path_tracing.cpp & brdf.cpp)
  • Explicit light sampling.
  • BSDF importance sampling (Both of cosine-weighted hemisphere distribution and GGX distribution).
• Physically-based BRDF (Cook-Torrance Model). (brdf.cpp)
• BVH Accelerated. (bvh/AABBTree.cpp & bvh/AABB.cpp)
• Texture mapping. (material/texture.h)
• Normal mapping. (path_tracing.cpp)
• Anti-alising by dithering sensor pixels. (Camera.cpp)
• Gamma correction. (main.cpp & reader/gamma_transform.h)
• Loading objects and materials from OBJ and MTL files. (reader/read_obj.cpp)

Improvements

• Parallelized with OpenMP.
• Better Polymorphism architecture.
  • Eliminated all dynamic_cast.
• Performance improvements.
  • Optimized ray-object intersection by pre-computing data and Möller–Trumbore intersection algorithm.
  • Replaced std::shared_ptr with std::unique_ptr or raw pointer where possible.
  • Replaced 64-bit double with 32-bit float for floating point data.
  • Utilized Eigen’s vectorized operations and lazy evaluation where possible.
• Modernized the code based on C++20 standard.

Rendered Images

Textured Materials

Comparing Parametric Materials

	Metallic = 0.0	Metallic = 0.5	Metallic = 1.0
Roughness = 0.1
Roughness = 0.5
Roughness = 1.0

Comparing Samples Per Pixel (SPP)

SPP = 1	SPP = 4	SPP = 16
SPP = 64	SPP = 256	SPP = 1024

Acknowledgements

Libraries

• Eigen
• OpenMP: For parallelization.
• libigl: For loading STL files.
• tinyobjloader: For loading OBJ and MTL files.
• stb: For reading and writing images.
• json: For parsing JSON files.

Assets

• Morgan McGuire’s Computer Graphics Archive
• Poly Haven
• Blender

Materials

Monte Carlo path tracing: https://sites.cs.ucsb.edu/~lingqi/teaching/resources/GAMES101_Lecture_15.pdf

Multiple Importance Sampling: https://computergraphics.stackexchange.com/questions/5152/progressive-path-tracing-with-explicit-light-sampling

Cook-Torrance BRDF: https://learnopengl.com/PBR/Theory

Cosine-weighted Importance Sampling: https://ameye.dev/notes/sampling-the-hemisphere/

Importance Sampling for GGX Distribution:

• https://agraphicsguynotes.com/posts/sample_microfacet_brdf/
• https://schuttejoe.github.io/post/ggximportancesamplingpart1/

Möller–Trumbore intersection algorithm: https://en.wikipedia.org/wiki/M%C3%B6ller%E2%80%93Trumbore_intersection_algorithm

Normal Mapping: https://learnopengl.com/Advanced-Lighting/Normal-Mapping