Survey of Markov Chain Monte Carlo Methods in Light Transport Simulation.

Area of Science:

• Computer Graphics and Rendering
• Computational Physics
• Statistical Computing

Background:

• Markov chain Monte Carlo (MCMC) methods were introduced to light transport simulation two decades ago.
• A significant body of research has followed, yet no comprehensive survey exists.
• Existing MCMC algorithms for light transport simulation address various challenges in rendering complex scenes.

Purpose of the Study:

• To provide the first comprehensive survey of Markov chain Monte Carlo (MCMC) algorithms applied to light transport simulation.
• To categorize existing MCMC methods based on their objectives and properties.
• To analyze the strengths and weaknesses of different MCMC approaches in rendering.

Main Methods:

• Categorization of MCMC algorithms by their specific goals and characteristics within light transport simulation.
• Analysis of how different MCMC techniques address core challenges such as convergence and efficiency.
• Inclusion of an introductory section on general MCMC principles with a simple illustrative example.

Main Results:

• A structured overview and classification of numerous MCMC algorithms developed over the past 20 years for light transport simulation.
• Identification of the key strengths and limitations associated with various MCMC strategies.
• Discussion on the common issues encountered in MCMC applications for rendering and potential solutions.

Conclusions:

• This survey offers a foundational resource for understanding the landscape of MCMC in light transport simulation.
• It highlights areas for future research, including method combination and performance enhancement.
• The paper aims to make MCMC methods more accessible to researchers and practitioners in computer graphics.