Progressive Lightcuts for GPU

Lightcuts [Walter et al. 2005] is an attractive rendering algorithm that efficiently handles complex lighting by adaptively approximating the illumination using clusters of virtual point light (VPL) sources. Two of its limitations are the infinite memory footprint required for variance convergence and the bias introduced by VPL contribution clamping. We present a progressive Lightcuts algorithm, which is consistent and converges using a fixed amount of intermediate memory. This is essential for high quality rendering, especially considering the tight memory budget on the GPU.