Position-Dependent Importance Sampling of Light Field Luminaires

Area of Science:

• Computer Graphics
• Computational Imaging

Background:

• Physically realistic image synthesis benefits from using real-world light sources (luminaires).
• Light field models offer accurate near-field representation and direct acquisition capabilities.

Purpose of the Study:

• To introduce a dynamic sampling strategy for complex light field luminaires.
• To develop an unbiased estimator for efficient and realistic image synthesis.

Main Methods:

• Dynamic adaptation of sampling domains for each 3D scene position and frame.
• Balancing sample distribution across representation components using affine transformations and CDF restrictions.
• GPU-based implementation for efficient computation.

Main Results:

• The dynamic sampling strategy significantly reduces the number of samples required for converged results.
• Achieved interactive frame rates for direct lighting with large measured luminaires.
• Demonstrated efficiency on modern hardware with a GPU implementation.

Conclusions:

• The proposed dynamic sampling strategy enhances the efficiency of light field rendering for realistic image synthesis.
• The method enables interactive rendering of complex scenes lit by real-world luminaires.