Direct Transmittance Estimation in Heterogeneous Participating Media Using Approximated Taylor Expansions

Area of Science:

• Computer Graphics
• Computational Physics
• Numerical Analysis

Background:

• Accurate light transport simulation in participating media is crucial for realistic rendering.
• Existing transmittance estimation methods lack theoretical understanding and can be computationally intensive.
• Evaluating transmittance involves intractable integrals of extinction coefficients.

Purpose of the Study:

• To develop a new class of unbiased transmittance estimators.
• To provide theoretical grounding for transmittance estimation algorithms.
• To improve the efficiency and accuracy of light transport simulations.

Main Methods:

• Introduced unbiased transmittance estimators based on Taylor expansion truncation and random sampling.
• Developed estimators using importance sampling and Russian roulette for finite results.
• Presented non-analogous methods that directly sample extinction functions.

Main Results:

• Demonstrated that new estimators are unbiased.
• Showcased improved performance over traditional algorithms for heterogeneous media.
• Validated the effectiveness of the new class of estimators on CPU and GPU.

Conclusions:

• The proposed estimators offer a theoretically sound and efficient alternative for light transport calculations.
• These methods advance the field of rendering and scientific visualization.
• Directly sampling extinction functions provides significant performance gains.