(L) [2018/07/13] [ost
by b_old] [Decomposition tracking question] Wayback!I have a working implementation of spectral tracking as described in [LINK http://drz.disneyresearch.com/~jnovak/publications/SDTracking/SDTracking.pdf this paper] and am trying to combine it with decomposition tracking, also described in that paper.
I decided to implement the simplest version first, ignoring weights and spectral data. The source code of my attempt can be found in this [LINK https://gist.github.com/Opioid/43e695cf5ad88a3e3243c97c53c13823 gist]. It is supposed to be a direct translation of the AnalogDecompositionTracking() function from the paper.
My test case is an achromatic, homogenous volume which I have artificially divided in control component and residual component, so that both together add up to the total collision coefficients. My decomposition tracking implementation fails with this simple setup.
Intuitively I thought I could simply run two delta tracking functions, one for the control component and one for the residual component and in combination this would lead to the correct result, as with calculating transmission. This also did not work.
Can you see where things go wrong?