Photon beams have a significant performance and quality benefit over standard volumetric photon mapping, while requiring less photons. This representation can also be viewed as a way to implicitly increase the effective "resolution" of a photon map, thereby reducing bias (blur) and also significantly reducing variance.
Photon beams have a significant performance and quality benefit over standard volumetric photon mapping, while requiring less photons. This representation can also be viewed as a way to implicitly increase the effective "resolution" of a photon map, thereby reducing bias (blur) and also significantly reducing variance.
Photon beams have a significant performance and quality benefit over standard volumetric photon mapping, while requiring less photons. This representation can also be viewed as a way to implicitly increase the effective "resolution" of a photon map, thereby reducing bias (blur) and also significantly reducing variance.
We have developed a novel, comprehensive theory of volumetric
radiance estimation. This theory allows for estimating in-scattered
radiance at a point, or the accumulated in-scattered radiance along a camera beam. Both of these operations can be performed using the standard photon map representation as done in previous work. Moreover, our theory generalizes both of these operations by introducing the concept of photon beams, a more compact, and expressive intermediate representation of lighting in participating media. The combination of these two data representations and two query operations results in a collection of nine distinct radiance estimates for computing complex lighting in participating media. Due to the increased expressiveness of this new lighting representation, photon beams have a significant performance and quality benefit over standard volumetric photon mapping, while requiring less photons. This representation can also be viewed as a way to implicitly increase the effective resolution of a photon map, thereby reducing bias (blur) and also significantly reducing variance. Using this representation, we are able to render extremely sharp details