What colour temperature is subjectively closest to "pure white"?

[fyrstormer]

Not every space needs to be a lit like a laboratory.

Nice example.

"Pure white" is in the vicinity of 6000K, but as StarHalo pointed out, not every situation calls for pure white light. Personally I prefer to stay in the range of 4500K-7500K (which is roughly the range from late-afternoon sunlight to rainy-day "cloudlight") because going outside of that range hurts my eyes if the light is being used for general illumination. Needless to say, I was not a big fan of the first generation of HID headlights, but fortunately they've gotten a lot better since then.

In situations where the ambient lighting is going to be dim, it's generally a good idea for the tint to be towards the cooler end of the spectrum, because the human eye isn't really designed to work with small amounts of yellow light and nothing else -- it is designed to work with small amounts of blue light and nothing else, but not the other way around. I've actually been in restaurants before where the lighting was so "intimate" (i.e. illuminated with nite-lite bulbs, or close to it) that I got a headache just trying to read the menu and I couldn't wait to leave. Those places were overdoing the "warm intimate lighting" effect, but within a reasonable range of brightness, warmer light is definitely more "cozy" even if not as chromatically accurate.

 

[fyrstormer]

Right, if you're in a 3000K room, a 4500K light will look pure white. The problem is, once you move to a 6500K room, that same light will now look warm.

That's why it's better (for the purposes of the OP's question) to evaluate ambient lighting itself, instead of spot-lighting inside an otherwise ambient-lit space. And in that context, 6000K is going to look like pure white, because that's what noon sunlight is, and I think we all can agree that the Sun is the ultimate ambient light source.

I have a couple flashlights with 4500K emitters in them and they always make plants look dead and wilted when I shine it on them at night. The factoid about warm-tinted LEDs neglecting the green component explains that pretty well; I'm looking forward to my forthcoming batch of Hi-CRI emitters so I can see how they look by comparison.

 

[fyrstormer]

Humans have a built-in color balance in their brains. As a result, a fairly wide range of color temperatures look "white" once you get used to them. The real question should be what color temperature looks white to us without this automatic color balance? Speaking from experience, this is likely to be the most pleasant color temperature to live under because our brains won't be constantly trying to correct for white.

Yes, and if you pay close attention to what you're seeing, you can tell when you're compensating for the off-whiteness of the ambient lighting. I have a halogen lamp next to my desk that looks very warm compared to the fluorescents overhead, but when I look at a piece of paper on my desk that's being lit by the fluorescents, it's still apparent to me that the reflected light is somewhat warmer than "pure white". It's close enough that I can choose to ignore the difference if I want to, but I can perceive the difference nonetheless.

 

[Twinbee]

And in that context, 6000K is going to look like pure white, because that's what noon sunlight is,

I'd be wary about using the sun as a yardstick for white for a few reasons (colour temp changes throughout the day, the rest of the sky to consider, and because what we perceive as nearest true white may not be completely based on the sky/sun anyway).

I think it would be safer (though still dangerous) to base 'true white' off all visible wavelengths consisting of equal amplitude. I wonder what colour temperature that would be closest to...

 

[fyrstormer]

What other yardstick could you possibly use? The Sun has been providing illumination for us since before we figured out how to build campfires.

The color temperature of direct sunlight does change throughout the day, hence why I specified noon sunlight, but the overall color temperature of direct+indirect sunlight doesn't change until the sun is low enough that some of the light gets scattered back into space from the low angle of incidence.

 

[Twinbee]

Well, I've given two already. In summary:

1: Acclimatize test subjects to pitch black for about 10 minutes to an hour. Then let them control a wheel which adjusts the colour temperature of a light and ask them to pick what they think is nearest to true white. Get them to look at the light (dim it as appropriate). Average the results of many people and call that 'white'. I think this is the safest way.

2: Produce a light which has equal amounts of power across the whole of the visible spectrum, and call that white. This is an interesting approach, but also dangerous. It would be interesting to see if the colour temp matches (1).

 

[jtr1962]

I think it would be safer (though still dangerous) to base 'true white' off all visible wavelengths consisting of equal amplitude. I wonder what colour temperature that would be closest to...

That would be equal energy white, or illuminant E, and it has a CCT of ~5455K ( note that illuminant E is off the Planckian locus ).

 

[Twinbee]

Interesting thanks. Yes, Illuminant E is slightly more magenta/purplish than anything an incandescent light source could produce.

Another idea of representing 'true white' is the level at which the three eye cone types are equally stimulated. As far as I know, light which stimulates the eye's 'blue' colour cone also stimulates the 'red' colour cone slightly, so there may be some compensation needed.

However, even then I'd still prefer to class 'true white' as Illuminant E, and say that we all have faulty eyes

 

[Lawliet]

Coming from a different angle:
Photography has the discharge of a flash tube as a de facto standard. You'd be hard pressed to find a flash in working condition thats more than 100-150k from 5600k.
To keep logistics at bay daylight films where balanced for just this color temp, the same goes for conversion filters.
Go a step further, those flashes are the light source that was used to take all the pictures in catalogs and ads, telling us how the product is supposed to look.

Maybe not exactly scientific, but working practice.

 

[Twinbee]

Interesting too. I wonder what the CRI of that flash is and how close it is to a flat spectrum.

 

[Lawliet]

They are about half black body radiation, half a blend of red, green and blue spectral lines that get smeared because of the high pressure/temperature.

I haven't seen CRI measurements, but the only time someone makes a fuss about lighting selection is when some pigments are UV active.:thumbsup: Thats from people who complain if you use incans of different age in one lighting setup.

 

[Kinnza]

The problem with this question is each person has a different expectation about what "true white" should be. So the first consensus about it is that there is no universal "true white", neither from an objective of subjective viewpoint.

Color perception varies from person to person, although fortunately it follows some universal paths. But the main complication is color is a multi dimensional concept. Should be "true white" about hue? In this case, a pure white should fall on the planckian locus or very close to it, say no more than 3 step McAdam ellipse and preferably, for definition, into a 1 step one (Duv<0,0007) so no any hue ("tint") is noticeable. I think in this point most people agrees. And between 3500 and 7000K.

But color has many more dimensions, and agreeing which of them is more important is way more subjective. It depends of personal preference and somewhat on the culture you have grown in. Should "true white" render colors as more similar as possible as sunlight? If so, what sunlight phase? Morning, noon, afternoon? Cloudy or clear sky? In general, most consensus along the color scientific community has been to choose sunlight like light sources, the most prominent the Illuminant D65, in which is based CIELAB, the main official (CIE) color space. But many people feels 6507K as too "cold".

It seems most people favor other dimension of color and light: its effect on our mood. Almost perfect white sources, as D65, but with a high percentage of blue white may be perfect for many people on a working space or when they want a feeling of cleanliness, but too "active" for a living room, where people prefer warmer tones so they feel more relaxed.

Another dimension of lighting is the level of light itself. This parameter is often overlooked, but it is very important. A same color of light has different effect on our mood and render colors differently as luminance level varies. Although our brain usually compensates for changes and contrast when luminance varies, it is not the same a D50 at 5 cd/m2 than at 100 cd/m2. Actually, I think many people that identifies incandescent lamps as the most pleasant white would be surprised if tries a colder tone, as 4500K but at lower luminance.

Should "true white" be pleasant or reliable on color rendering? I think the point most difficult to reach a consensus is this. Should it make scenes appear "natural" (whatever it means) or vivid, with enhanced color?

In my personal preference, I think a true white is somewhere between 4000 and 5000K, very close to planckian locus and providing a luminance matched to each application.

 

[smarkum]

Reviving a sleepy thread. . . this caught my interest as I spend 8-10 hours a day in an office lit by 3 lamps. Initially I just put the reg. bulbs in there that were "full spectrum" 40 or 60 watt. The result was flickering . . . I work in a big office buliding so checking for the power flow to the outlets is out. So, I then tried those low watt cfl bulbs that were rated "warm". The flickering was much better , but still there. I'm VERY aware of such flickerings, a sad thing for sure! Finally I have in there two LED bulbs that use 6 watts or something like that. I'm at home now, so I cannot confim what temp the bulbs are , but I know they are warm. My goal in my office was to create warm ambiance. . . and soft lighting. I did accomplish the soft lighting part, but after reading this thread, I think I've gone overboard on the warmth scale. And, I do get headaches nearly everyday at work (I thought it was due to the nature of my work). And, I find that at certain times it seems as though the room is "foggy". I know that probably doesn't make sense, but that is how it seems by the end of the day and when my eyes and ears are REALLY tired. So I think I will get some bulbs today that are in the 4500 - 5500 range and see what diff. that makes. This has been an expensive experiment. And, considering the LED are the way I must go for less flickering, it will be even more expensive after the new bulb purchase. If anyone has any thoughts on my approach, I'd love to hear them. And, thanks for this most informative thread!

 

[fyrstormer]

Use a halogen desk lamp. The rest of the office can be lit by flickery fluorescent bulbs and it won't matter as long as the light near your workspace is steady.

 

[smarkum]

good info to have . . . thanks fyrstormer!

 

[felixnauta]

Re: What colour temperature is subjectively "pure white"?

My testing shows 4300K to be the mark, but 4200-4600 is close enough in most applications. 5000K is still pretty blue. 3000K is quite yellow.

One of the issues is that this scale is designed from the use of a single technology (glowing hot metal). So a non-glowing-metal source (LEDs for example) can be (measure at exactly) 4300K and still be tinted with a color not on the color temp scale, like green.

I agree with you because the question in the thread's title says: "subjectively".

And by the other hand we must to remember that "daylight" in really fact means a narrow but variety range of tone colors: at the morning tends a bit to blue; and in the evenings turns a bit to the yellow.

But the word is: "pure white" I am waiting for the arrival of some cfl's; then I will test for myself; some facts about some labels in the lamps.

 

[markr6]

Re: What colour temperature is subjectively "pure white"?

What really confuses me is that the 4100K T8 fluorescents in my home closets are nice and white, (but I would say too cool for living spaces). But my Zebralight neutrals 4200K and 4400K lights are completely different. If my closet lights are really 4100K, the zebralights have to be below 3000K. I couldn't imagine the tolerance to be so high.

My Fenix PD32UE torches are also very close to pure white, I'm guessing 5000K. But still a little bluish.

L10s w/ Nichia 219, PURE white (high CRI or not)

 

[ElectronGuru]

Re: What colour temperature is subjectively "pure white"?

What really confuses me is that the 4100K T8 fluorescents...

The problem is the scale itself. Color temp is linear, 1 dimension like a long thin road. Designed to measure incan sources, its like stating your location in a small town based solely on how far along a central railroad track you are standing.

Modern light sources are 2 dimensional, having multiple colors and other variations. A given LED or fluorescent light source can be so far away from that center, as to be meaningless. They K value is simply how close you can get to actual, on that railroad track. Distance from the track is completely ignored.

 

[markr6]

Re: What colour temperature is subjectively "pure white"?

The problem is the scale itself. Color temp is linear, 1 dimension like a long thin road. Designed to measure incan sources, its like stating your location in a small town based solely on how far along a central railroad track you are standing.

Modern light sources are 2 dimensional, having multiple colors and other variations. A given LED or fluorescent light source can be so far away from that center, as to be meaningless. They K value is simply how close you can get to actual, on that railroad track. Distance from the track is completely ignored.

:twothumbs Got it...thanks!

 

[fyrstormer]

Re: What colour temperature is subjectively "pure white"?

The distance away from the actual blackbody radiation line-graph isn't ignored, it's expressed as Color Rendering Index. 100% CRI means the light contains the exact radiation spectrum indicated by the color-temperature it is associated with. I don't know if 0% CRI exists, but I suppose it would mean the light has the same tint as the color-temperature it's associated with, but it's all the way at the edge of the visible-color gamut, far away from the blackbody radiation line-graph.

One example would be cyan light, which is a single color, but it *could* be labeled as 40000K with 0% CRI -- at a casual glance the overall tint of the light source might look the same, but the cyan light would be completely incapable of rendering any colors except cyan, unlike true 40000K light, which can render all visible colors, albeit with a strong bluish tint.

On the other hand, a deep red light would also be single-color, but could be labeled as 1000K with 100% CRI. That's because the dimmest blackbody radiation that people can see only consists of a narrow range of deep red light, mixed with a bunch of infra-red that we can't see, so a single-color deep red light would have the same rendering capability as true 1000K light.

If there's a problem with using blackbody radiation to rate the quality of light, it's mostly that people don't actually know what makes good light anyway, so they use hot glowing things as a benchmark, because they can instinctively relate to it.