I thought that this concept was really interesting as I didn't know how different spectra can produce the same colors in our eyes. I wonder if there are any underlying differences or advantages to using one spectrum as opposed to another.
ttalati
Is the reason that metamerism exists is because at a high level we are converting an infinite domain space into a small scalar value (or I guess 3 values that get weighted to lead to one value)? I imagine with such condensement of information, it builds this redundancy that we perceive as metamerism
rcorona
@ttalati, if I understand correctly I think that interpretation is correct.
The original space is infinite dimensional since you could think of any unique wavelength as a dimension, and so there are infinitely many since the space of wavelengths is real-valued.
The signal for a spectral distribution is then projected into 3 scalar responses for the 3 different types of cones.
This reduction in dimensionality means that there are many "collisions" between different points in the infinite-dimensional space since their projection results in the same 3D coordinate.
rcorona
As a follow-up, I'm curious if a measure exists for how much redundancy or how many "collisions" there are as a result of this dimensionality reduction.
In other words, given the distribution of naturally occurring spectral energy distributions, can one map out which regions in the space have more/less metamerism than others?
I thought that this concept was really interesting as I didn't know how different spectra can produce the same colors in our eyes. I wonder if there are any underlying differences or advantages to using one spectrum as opposed to another.
Is the reason that metamerism exists is because at a high level we are converting an infinite domain space into a small scalar value (or I guess 3 values that get weighted to lead to one value)? I imagine with such condensement of information, it builds this redundancy that we perceive as metamerism
@ttalati, if I understand correctly I think that interpretation is correct.
The original space is infinite dimensional since you could think of any unique wavelength as a dimension, and so there are infinitely many since the space of wavelengths is real-valued.
The signal for a spectral distribution is then projected into 3 scalar responses for the 3 different types of cones.
This reduction in dimensionality means that there are many "collisions" between different points in the infinite-dimensional space since their projection results in the same 3D coordinate.
As a follow-up, I'm curious if a measure exists for how much redundancy or how many "collisions" there are as a result of this dimensionality reduction.
In other words, given the distribution of naturally occurring spectral energy distributions, can one map out which regions in the space have more/less metamerism than others?