Am 01.12.2017 um 20:42 schrieb Sven Littkowski:
> The images still look nice. Sad only, that I self have no understanding
> of media, I wished I could help here. Sounds like you should propose a
> new feature inside the BETA section of this newsgroup, or expanding
> existing features to make it possible to increase the density further.

The issue isn't so much whether it is /possible/ to increase the
density, but rather whether doing so gives /realistic results/.

The problem is that the scattering media model is designed for cases
where the media's density is non-uniform, but low enough that the single
scattering contribution is dominant (i.e. a typical light ray will be
deflected from its initial direction at most once), and the multiple
scattering contribution (i.e. light that is deflected from its initial
direction more than once) is negligible.

Compare this to the subsurface light transport model, which is
specifically designed for cases where the density is pretty much
uniform, but high enough that the multiple scattering contribution is
dominant.


To extend the scattering media model in such a manner that it gives
reasonably realistic results for high-density media, a monte-carlo-ish
rendering approach would have to be used, where secondary rays would be
shot here and there along the light's path (even recursively). Expect
noisy results and/or high rendering times.

This is in fact on my wishlist, but other stuff takes priority.

Post a reply to this message