In spectroscopy, we generally look at an object of some composite color (ie a white light, a red cuvette of nanoparticles, a green lawn) and separate the composite colors into a spectrum. For example, the transmission through an object as a continuous function of wavelength, T(lambda).
What I'd like to know is how can we reverse this process. IE, given a predicted spectrum T(lambdas), how can I estimate what the overall color of this object will be in the visible wavelengths? IE, how to go from T(lambda) to a single, 3-channel rgb color that I can print on a computer screen?
Here's an example with gold, that shows the relation to hues of red corresponding to absorption spectra:
http://people.csail.mit.edu/jaffer/FreeSnell/granular.html#Gold Ruby Glass
Thanks.
Hi,
it should be quite easy - you multiply the given spectrum with R/G/B spectral sensitivity curves. It means - you take the intensity value of given spectrum at each wavelength and multiply it to the sensitivity value of each curve at this wavelength. Then you integrate the areas of resulting curves (just sum the multiplication products) - these three numbers give you the ratio of R/G/B.
You should care though that R/G/B spectral sensitivity curves of human eye (e.g. http://en.wikipedia.org/wiki/Spectral_sensitivity) is somewhat different from those of R/G/B sensors. And that for accurate color representation on computer screen its spectral emission curves should be taken into account (search for color calibration).
Hello Adam,
You might find the webpage "Colour Rendering of Spectra" to be useful (link below). A program in C is included for calculations from absorbance spectra. I have never used it myself, but it looks like what you are looking for.
Good luck!
https://www.fourmilab.ch/documents/specrend/
Thanks guys. I've never ever heard of a spectral sensitivity curve, but that's wild!
- Badges
- Science topic