So I'm reading a computer graphics research paper and I'm confused as to how to interpret certain formulas that are being used.
In the paper, a value σ is defined as an RGB triplet. Later on, that same triplet is used in the calculation of scalar values. For instance, a floating point value S is defined using the following equation:
float S = -log(ξ)/σ, where ξ is a random variable between [0,1].
I see this kind of thing come up a lot in computer graphics papers, not just this one, and I don't really know how to interpret it. Do I take the average of the 3 RGB values? Do I use the luminance value? Do I add them all up?
If anyone here has experience with this, I appreciate your input.
Edit: Here is the link to the paper - http://www.graphics.stanford.edu/papers/bssrdf/bssrdf.pdf
On page 2, σt is defined as σa + σs.
On page 5, you can see that σa and σs values are defined in terms of RGB. This means that σt is also an RGB triplet.
Then on page 6, they say this: "We pick a random distance, s'o = log(ξ) / σt.
Distance is a scalar quantity, so how does this equation work?
Up to my understanding, $\sigma$ are not triplets. Instead, they are functions of the frequency $\omega$ of light over the visible spectrum. These $\sigma(\omega)$ are really infinite dimensional instead of 3 dimensional objects.
For ordinary material, if you shoot a monochromatic laser beam with a fixed frequency $\omega$ to it, it will reflect/absorb/scatter light back with exactly the same frequency $\omega$. The corresponding physical color won't change. This means one can analysis the problem for each frequency channel independently. Within each frequency channel, all $\sigma(\omega)$ behave like scalars.
Of course, human eyes has only 3 types of cells to detect colors. So biologically, we will approximate all the physical colors by a 3-dimensional RGB triplets. For application in the computer graphics, one usually just take the lowest order approximation. You treat these $\sigma$ as a 3-dimensional vector of coefficients. All formula will be interpreted as one operating on the three color channels independently. i.e
$$S = -\frac{\log\xi}{\sigma} \quad\iff\quad \begin{cases} S_R &= -\frac{\log\xi_R}{\sigma_R}\\ S_G &= -\frac{\log\xi_G}{\sigma_G}\\ S_B &= -\frac{\log\xi_B}{\sigma_B} \end{cases}$$