Prism Spectrum for Color Calibration

I'm wondering, why use these expensive color charts for color calibration, when you could get the real thing from a prism.

Tom Axford wrote:

SmoothOperator wrote:

I'm wondering, why use these expensive color charts for color calibration, when you could get the real thing from a prism.

There are infinitely many more colours than those in the spectrum. Most real colours are mixtures (e.g. browns) and do not occur in the spectrum which consists only of "pure" colours.

The are other colors yes but not many more. The spectrum consists of a near infinite variety of subtly changing colors. If you start investigating atomic spectroscopy you'll soon see just how many colours there really are. Humans can just about distinguish colors with a wavelength difference of 2nm giving around 200 colors in the spectrum. The same appearance of color can be achieved by mixing the correct proportions of primary colors.

The trouble with using a prism for your color calibration is that you don't have it calibrated at all you just get a band of varying color, and with a camera will often only see  relatively few distinct changes. Color charts have carefully chosen shades which are also separated to make them easier to tell apart.

FWIW I have used diffraction gratings to investigate the spectral distribution of lighting with my full spectrum camera (where the actual wavelengths/colors the camera responds to  stretches from 350nm to 1100nm rather than just the typical visual range around 400nm to 700nm. To get any useful data I would have to be very repeatable in positioning & make use of a range of optical filters. Fortunately the UV/Vis/NIR spectrometer at work allows me to measure absorbtion/transmission far more accurately them my photography can use (to better than 1nm wavelength & <0.1% transmission) but it only works using it's built in light sources. (The atomic spectrometer has at least a thousand fold better wavelength control but isn't practical for scanning wide ranges and can't hold useful items like lenses instead needing it's samples vapourised)

Tom Axford wrote:

petrochemist wrote:

Tom Axford wrote:

SmoothOperator wrote:

I'm wondering, why use these expensive color charts for color calibration, when you could get the real thing from a prism.

There are infinitely many more colours than those in the spectrum. Most real colours are mixtures (e.g. browns) and do not occur in the spectrum which consists only of "pure" colours.

The are other colors yes but not many more. The spectrum consists of a near infinite variety of subtly changing colors.

You have already contradicted yourself in just two sentences!

Actually the main contradiction is in just two words 'near infinite'

The wavelength of a photon is quantized so the number of colors between two wavelengths is not actually infinite, but the steps are very small. Hence the meaning less phrase 'near infinite' to imply a huge number. Visual light is roughly 400-700nm. I've come across wavelengths specified to 4 or more decimal places, (such as the sodium D lines at 588.9950 & 589.5924nm ) so at the resolution of the data I've seen that's ~3million pure colors - I believe the quantum steps in wavelength are actually considerably smaller variations, and I've never come across or heard of an instrument capable of resolving them.

Tom Axford wrote:

petrochemist wrote:

Tom Axford wrote:

petrochemist wrote:

Tom Axford wrote:

SmoothOperator wrote:

I'm wondering, why use these expensive color charts for color calibration, when you could get the real thing from a prism.

There are infinitely many more colours than those in the spectrum. Most real colours are mixtures (e.g. browns) and do not occur in the spectrum which consists only of "pure" colours.

The are other colors yes but not many more. The spectrum consists of a near infinite variety of subtly changing colors.

You have already contradicted yourself in just two sentences!

Actually the main contradiction is in just two words 'near infinite'

The wavelength of a photon is quantized so the number of colors between two wavelengths is not actually infinite, but the steps are very small. Hence the meaning less phrase 'near infinite' to imply a huge number. Visual light is roughly 400-700nm. I've come across wavelengths specified to 4 or more decimal places, (such as the sodium D lines at 588.9950 & 589.5924nm ) so at the resolution of the data I've seen that's ~3million pure colors - I believe the quantum steps in wavelength are actually considerably smaller variations, and I've never come across or heard of an instrument capable of resolving them.

I have never heard of a 'quantum of wavelength' before, but if such exists, what value does it take? Perhaps the Planck length, which is about 10^(-35) metres? On that assumption, there are well over 10^20 separate wavelengths within the visible spectrum!

I think this is all very much in the realm of speculation. I've never heard of any experimental confirmation of this. Are you aware of any?

Sounds about right hence my 'near infinite'. No I've never come across experimental confirmation. If the theory is wrong & wavelength is a true continuum then there are an infinite number of wavelengths & an infinite number of pure colors.

Of course neither commercial digital cameras or the human eye can resolve anywhere near this number of colors. Sensors in some astronomical gear can resolve impressive numbers (as produced in their stellar spectra) but they're not exactly pocketable systems

SmoothOperator wrote:

petrochemist wrote:

Tom Axford wrote:

SmoothOperator wrote:

I'm wondering, why use these expensive color charts for color calibration, when you could get the real thing from a prism.

There are infinitely many more colours than those in the spectrum. Most real colours are mixtures (e.g. browns) and do not occur in the spectrum which consists only of "pure" colours.

The are other colors yes but not many more. The spectrum consists of a near infinite variety of subtly changing colors. If you start investigating atomic spectroscopy you'll soon see just how many colours there really are. Humans can just about distinguish colors with a wavelength difference of 2nm giving around 200 colors in the spectrum. The same appearance of color can be achieved by mixing the correct proportions of primary colors.

The trouble with using a prism for your color calibration is that you don't have it calibrated at all you just get a band of varying color, and with a camera will often only see relatively few distinct changes. Color charts have carefully chosen shades which are also separated to make them easier to tell apart.

FWIW I have used diffraction gratings to investigate the spectral distribution of lighting with my full spectrum camera (where the actual wavelengths/colors the camera responds to stretches from 350nm to 1100nm rather than just the typical visual range around 400nm to 700nm. To get any useful data I would have to be very repeatable in positioning & make use of a range of optical filters. Fortunately the UV/Vis/NIR spectrometer at work allows me to measure absorbtion/transmission far more accurately them my photography can use (to better than 1nm wavelength & <0.1% transmission) but it only works using it's built in light sources. (The atomic spectrometer has at least a thousand fold better wavelength control but isn't practical for scanning wide ranges and can't hold useful items like lenses instead needing it's samples vapourised)

It just seems that the issues of having a calibration of the spectrum from a prism wouldn't be difficult for an entrepreneur to work through. For example a device could be fabricated such that a slit of white light is directed through a prism then through some sort of bar coded pattern that could then be read by appropriate software that would pick up the pattern of dots in the image. The pattern of dots would be based on theoretical scattering angles, and pre calibrated as a gauge. Seems like this could be done for less cost than the pigments that go into that color checking card, and would have the potential to test more colors...

Some of what you're describing (and possibly exactly what you're thinking of) seems to be along the lines of Iliah Borg's work on Spectron - https://www.dpreview.com/forums/post/61332698