The author describes color rendering index as a percentage; it is not. It's 100 - [a calculation of average color error on 8 color samples]. Reporting the color error directly would probably be better; 90% and 95% would both be pretty good grades on most tests, but a rendering error of 10 sounds a lot worse than a rendering error of 5.
Being based on only 8 samples, CRI Ra doesn't tell the whole story. There are other standards like TLCI that use more samples, and six supplemental samples some sources report. R9 (strong red) and R12 (strong blue) are particularly relevant to modern lighting, as LED sources often struggle with them despite high CRI (Ra) ratings.
CRI is a comparison to blackbody radiation (like an incandescent bulb) or simulated daylight at the same correlated color temperature as the sample light source. An incandescent bulb and midday sunlight at the equator both have CRI of 100, but a CCT of 2700K and a CCT of 5700K will render colors very differently. White balance in a camera's settings or raw development software can make one look like the other.
CCT is roughly blue-yellow balance, but light sources can deviate from blackbody radiation on a red-green axis. This is most often expressed as "delta u v" or "Duv", as it refers to distance in the CIELUV color space[0]. Duv numbers are small, with differences of less than 0.001 being noticeable in a side by side comparison and more than +/- 0.006 falling outside the ANSI standard for white light. Positive numbers are green, and negative numbers are red/pink.
I sort of wonder is there a "true" way to represent color? Would a intensity over frequency curve do it?
I probably lack the terminology to express myself properly, but color ends up being a very human thing, with a huge focus on primary colors(the colors that match the frequency response of the human eye) This is fine, we are human after all, but then I always feel color theory dives off the deep end, with color wheels and complimentary colors and triplets, etc. why? there is nothing physically cyclic going on. It never makes any sense to me that patterns based on it being cyclic would be any better than just picking colors. Any way, silly rant aside, if we wanted to define an objective physical color format, how would it be done?
> I sort of wonder is there a "true" way to represent color? Would an intensity over frequency curve do it?
There’s no single “true” way because color isn’t a fixed property of light itself, it’s the result of how a visual system interprets wavelengths.
Having said that, if you’re looking for an objective way that doesn’t depend on observer, you could use spectral power distribution.
That would be the mix of wavelengths present, plotted as intensity vs. wavelength.
That’s because a single wavelength (say 580nm) produces one specific color sensation, but many different combinations of wavelengths can produce that identical sensation (called metamers), so representing color by wavelength alone isn’t even uniquely “true” perceptually.
It’s actually even worse because, as countless color illusions demonstrate, our perception of color is a result of much more than just the spectral power distribution of the light.
You could go a long way by specifying the reflectiveness over the visible light spectrum and then do the same for light sources. That way, you could do GDR style truly monochrome sodium vapor lamps, which makes telling colors apart impossible.
To do neon colors, you would need some sort of frequency redirect in your texture data as in "this pixel absorbs 80% infrared, but reemits it as visible red". Then you could have black light light sources and cool poster textures as well.
I think doing textures in this format would take too much space, but doing a 256 color palette that way might be fine.
If you plot the three primary colors with all possible combinations of only two of them, you can connect them all neatly using a circle, and distance on that circle does tell us something about how visually distinct colors are. It isn't much more profound than that.
One of the colors on the circle, magenta, has no corresponding pure wavelength, because in order to see magenta, your red and blue cones must be activated, but not the green cones -- you need two frequency peaks to do this. If we had four different types of cones, like many birds, there would be many more non-spectral fully saturated colors and you'd need something like a hypercube to represent them.
That should work as a brute-force method. You could simulate lighting conditions by measuring the reflected spectrum of different materials and multiplying with the spectrum of a light source.
Check out the CIE XYZ color space and its color matching functions. It's the standard for defining/converting color spaces and calibrating displays for accurate color matching. It only models human perception, and contains colors that are physically impossible to reproduce.
Depending on what you want to do there's CIE XYZ (maps spectrum of color to a 3 component vector by integrating the spectrum multiplied by three functions), CIE Lab* (describes the color of objects under a specified illuminant), CIE Luv* (to characterize emissive source, nonlinear so the the distance between colors corresponds to perceived differences).
These color spaces are built to objectively describe colors, as seen by a human with typical color vision.
“I’ve thought about this for five minutes, and it seems so simple. It must be the tens of thousands of vision scientists and psychophysicists who have worked on this for the last century who have all missed how simple it is.”
CRI of lights is definitely a large factor in my purchasing these days. The muting, compressive response of some lights almost pains me at this point.
As this post's author noted white LEDs can be pretty solid. Bulbs that have a mix of white LEDs at various temperatures to be "bi-color" tend to be pretty solid as well. Its once you go towards RGB light bulbs that you tend to get a fairly iffy CRI.
Consider looking at TM-30 as well as CRI, it's a more useful metric that better takes R9 into account. I've gotten a little bit obsessed with colour in the last few months, and have been buying different bulbs for the purpose of making DCP profiles and doing film scanning. What you want, specifically, is violet pumped LEDs. They're expensive and it seems like there are only a handful of them on the market, but they're a lot more balanced spectrally and tend to have low flicker (probably just because they're a premium product).
> It's probably a good idea to buy a few samples of several different types and brands, test them all, and then decide which to buy in quantity for the home.
I'm looking forward to that. I own a bunch of Philips WhiteDial bulbs that are rated CRI90. It would be nice to test that and know how they really compare. No doubt there's a lot of dishonest CRI ratings out there.
There are no objective "basic colors". What "white" means is "excites our different types of photoreceptors roughly equally". And the photoreceptors have the sensitivity curves that they do because of our sun’s surface temperature and a convenient "window" in the opacity of our atmosphere. Organisms evolved under a different star, within a different atmosphere would have a different idea of what a "neutral" color is.
(Because of additional processing done by our visual system, what looks "white" to us is very dependent on the color of the ambient light. But the ~5800 K blackbody spectrum could be called the "most neutral" white to our eyes because that’s what evolution has optimized them for.)
The vision system has evolved to consider as white the average color of the environment.
In terrestrial environments, the average color is typically not much different from the true color of the Sun. The white light of the Sun is split by the atmosphere into the bluish light of the sky and the yellowish direct light, then those are reflected and diffused by various bodies until being collected by an eye or a photographic camera.
If the bodies in the environment are random enough, i.e. there are similar amounts of those which absorb or diffuse different parts of the spectrum (e.g. the environment includes brown earth, green plants, blue sky, variously colored flowers and fruits, etc.), then the average color of the light that reaches the photoreceptors will not differ much from the color of the original source of light, i.e. the Sun as seen from outside the atmosphere.
Sure, the sun is white, more or less (cf comments re human perception and what "white" really means).
But when did humans first perceive the sun as white? We can only reliably say "sometime on or after 12 April 1961", the day Gagarin went into space. Prior to that, as far as we know, all humans saw the sun as some other colour, most often yellow or red, depending on the time of day.
Since Newton's prism, we had pretty good reason for believing it was likely white, reasons that got stronger in the 19th C with improved theories of optics, the discovery of atomic spectra, etc. I'm curious to know when we confirmed for sure its whiteness, and whether this was before October 1957 (Sputnik) or after Gagarin or somewhere in between.
The direct light from the Sun is yellowish, because the blue light of the sky is subtracted from it.
If the sky is covered with clouds, then the light is white, because the blue sky light is mixed again with the direct Sun light.
If the sky is clear and there is an unobstructed view of it, the light reflected by a white object will be the sum of the light coming from the sky to the object and directly from the Sun to the object, so it will be white. For example this is true for white clouds and white snow.
This, i.e. that the Sun outside the atmosphere is perfectly white, has been understood for a long time, since the 19th century.
https://en.wikipedia.org/wiki/Black-body_radiation As objects get hotter and start glowing, they start to emit radiation in visible frequencies. Very hot objects keep glowing in the redder parts as they also start to glow in higher bands. Our eyes collect all the various frequencies at different intensities and experience them as one color. The second diagram on the page, with the chromaticity of different temperatures of ideal black-body radiators, shows the color that we perceive as the temperature of the glowing object goes up. The sun is around 5800k, so a little bit in in the yellow-green direction from perfectly balanced color response.
If you want to know why that diagram has its white point where it is, it's based on testing a bunch of people in laboratories. It's fairly predictable and repeatable, but it might not match your experience outdoors.
It's a convention. And at one time it was useful as the thing that could be replicated anywhere.
Also, things that are white can be judged by the amount of light they reflect or scatter as a percentage, independent of the source of illumination. For instance, among familiar objects, a common white ceramic bathroom tile comes pretty close to pure white.
Also, a thing can't be "whiter than white" in nature unless it's fluorescent. So it represents a kind of limiting case that people can agree upon.
The author describes color rendering index as a percentage; it is not. It's 100 - [a calculation of average color error on 8 color samples]. Reporting the color error directly would probably be better; 90% and 95% would both be pretty good grades on most tests, but a rendering error of 10 sounds a lot worse than a rendering error of 5.
Being based on only 8 samples, CRI Ra doesn't tell the whole story. There are other standards like TLCI that use more samples, and six supplemental samples some sources report. R9 (strong red) and R12 (strong blue) are particularly relevant to modern lighting, as LED sources often struggle with them despite high CRI (Ra) ratings.
CRI is a comparison to blackbody radiation (like an incandescent bulb) or simulated daylight at the same correlated color temperature as the sample light source. An incandescent bulb and midday sunlight at the equator both have CRI of 100, but a CCT of 2700K and a CCT of 5700K will render colors very differently. White balance in a camera's settings or raw development software can make one look like the other.
CCT is roughly blue-yellow balance, but light sources can deviate from blackbody radiation on a red-green axis. This is most often expressed as "delta u v" or "Duv", as it refers to distance in the CIELUV color space[0]. Duv numbers are small, with differences of less than 0.001 being noticeable in a side by side comparison and more than +/- 0.006 falling outside the ANSI standard for white light. Positive numbers are green, and negative numbers are red/pink.
[0] https://en.wikipedia.org/wiki/CIELUV
I probably lack the terminology to express myself properly, but color ends up being a very human thing, with a huge focus on primary colors(the colors that match the frequency response of the human eye) This is fine, we are human after all, but then I always feel color theory dives off the deep end, with color wheels and complimentary colors and triplets, etc. why? there is nothing physically cyclic going on. It never makes any sense to me that patterns based on it being cyclic would be any better than just picking colors. Any way, silly rant aside, if we wanted to define an objective physical color format, how would it be done?
There’s no single “true” way because color isn’t a fixed property of light itself, it’s the result of how a visual system interprets wavelengths.
Having said that, if you’re looking for an objective way that doesn’t depend on observer, you could use spectral power distribution.
That would be the mix of wavelengths present, plotted as intensity vs. wavelength.
That’s because a single wavelength (say 580nm) produces one specific color sensation, but many different combinations of wavelengths can produce that identical sensation (called metamers), so representing color by wavelength alone isn’t even uniquely “true” perceptually.
To do neon colors, you would need some sort of frequency redirect in your texture data as in "this pixel absorbs 80% infrared, but reemits it as visible red". Then you could have black light light sources and cool poster textures as well.
I think doing textures in this format would take too much space, but doing a 256 color palette that way might be fine.
https://en.wikipedia.org/wiki/Photon_upconversion
If you plot the three primary colors with all possible combinations of only two of them, you can connect them all neatly using a circle, and distance on that circle does tell us something about how visually distinct colors are. It isn't much more profound than that.
One of the colors on the circle, magenta, has no corresponding pure wavelength, because in order to see magenta, your red and blue cones must be activated, but not the green cones -- you need two frequency peaks to do this. If we had four different types of cones, like many birds, there would be many more non-spectral fully saturated colors and you'd need something like a hypercube to represent them.
That should work as a brute-force method. You could simulate lighting conditions by measuring the reflected spectrum of different materials and multiplying with the spectrum of a light source.
Check out the CIE XYZ color space and its color matching functions. It's the standard for defining/converting color spaces and calibrating displays for accurate color matching. It only models human perception, and contains colors that are physically impossible to reproduce.
https://en.wikipedia.org/wiki/CIE_1931_color_space
Depending on what you want to do there's CIE XYZ (maps spectrum of color to a 3 component vector by integrating the spectrum multiplied by three functions), CIE Lab* (describes the color of objects under a specified illuminant), CIE Luv* (to characterize emissive source, nonlinear so the the distance between colors corresponds to perceived differences).
These color spaces are built to objectively describe colors, as seen by a human with typical color vision.
As this post's author noted white LEDs can be pretty solid. Bulbs that have a mix of white LEDs at various temperatures to be "bi-color" tend to be pretty solid as well. Its once you go towards RGB light bulbs that you tend to get a fairly iffy CRI.
I've referenced https://optimizeyourbiology.com/light-bulb-database and https://optimizeyourbiology.com/smart-light-database before when looking at bulbs but if anyone has a better resource, absolutely please drop a reply.
Alternatively you buy a ColorChecker chart and measure the reflected light with a colorimeter.
Basically, the finer the color separation, wider the gamut, and whiter the light, the higher the CRI.
I'm looking forward to that. I own a bunch of Philips WhiteDial bulbs that are rated CRI90. It would be nice to test that and know how they really compare. No doubt there's a lot of dishonest CRI ratings out there.
(Because of additional processing done by our visual system, what looks "white" to us is very dependent on the color of the ambient light. But the ~5800 K blackbody spectrum could be called the "most neutral" white to our eyes because that’s what evolution has optimized them for.)
In terrestrial environments, the average color is typically not much different from the true color of the Sun. The white light of the Sun is split by the atmosphere into the bluish light of the sky and the yellowish direct light, then those are reflected and diffused by various bodies until being collected by an eye or a photographic camera.
If the bodies in the environment are random enough, i.e. there are similar amounts of those which absorb or diffuse different parts of the spectrum (e.g. the environment includes brown earth, green plants, blue sky, variously colored flowers and fruits, etc.), then the average color of the light that reaches the photoreceptors will not differ much from the color of the original source of light, i.e. the Sun as seen from outside the atmosphere.
But when did humans first perceive the sun as white? We can only reliably say "sometime on or after 12 April 1961", the day Gagarin went into space. Prior to that, as far as we know, all humans saw the sun as some other colour, most often yellow or red, depending on the time of day.
Since Newton's prism, we had pretty good reason for believing it was likely white, reasons that got stronger in the 19th C with improved theories of optics, the discovery of atomic spectra, etc. I'm curious to know when we confirmed for sure its whiteness, and whether this was before October 1957 (Sputnik) or after Gagarin or somewhere in between.
If the sky is covered with clouds, then the light is white, because the blue sky light is mixed again with the direct Sun light.
If the sky is clear and there is an unobstructed view of it, the light reflected by a white object will be the sum of the light coming from the sky to the object and directly from the Sun to the object, so it will be white. For example this is true for white clouds and white snow.
This, i.e. that the Sun outside the atmosphere is perfectly white, has been understood for a long time, since the 19th century.
If you want to know why that diagram has its white point where it is, it's based on testing a bunch of people in laboratories. It's fairly predictable and repeatable, but it might not match your experience outdoors.
Also, things that are white can be judged by the amount of light they reflect or scatter as a percentage, independent of the source of illumination. For instance, among familiar objects, a common white ceramic bathroom tile comes pretty close to pure white.
Also, a thing can't be "whiter than white" in nature unless it's fluorescent. So it represents a kind of limiting case that people can agree upon.