Consistent Post-Reconstruction for Progressive Photon Mapping

Hajin Choi, Bochang Moon

Gwangju Institute of Science and Technology

Accepted to Pacific Graphics 2021

Consistent Post-Reconstruction for Progressive Photon Mapping
Comparisons between two post-reconstruction techniques, single-buffered deep combiner (DC) [BHHM20] (c) and ours (d), which are integrated into stochastic progressive photon mapping (SPPM) [HJ09] (b). Both post-reconstruction techniques ((c) and (d)) effectively reduce the high-frequency noise in SPPM estimates, but our method produces sharper results than DC for the caustics (the bottom row). The number of iterations N_pass, where we use 0.1M photons per iteration, is adjusted so that each method uses approximately equal-render times, and we use relative mean-squared error (relMSE) [RKZ11] as a numerical measure.


Photon mapping is a light transport algorithm that simulates various rendering effects (e.g., caustics) robustly, and its progressive variants, progressive photon mapping (PPM) methods, can produce a biased but consistent rendering output. PPM estimates radiance using a kernel density estimation whose parameters (bandwidths) are adjusted progressively, and this refinement enables to reduce its estimation bias. Nonetheless, many iterations (and thus a large number of photons) are often required until PPM produces nearly converged estimates. This paper proposes a post-reconstruction that improves the performance of PPM by reducing residual errors in PPM estimates. Our key idea is to take multiple PPM estimates with multi-level correlation structures, and fuse the input images using a weight function trained by supervised learning with maintaining the consistency of PPM. We demonstrate that our technique boosts an existing PPM technique for various rendering scenes.