D65 and Daylight

[mcmorrison]

Hello,

As I understand it, D65 is a spectral curve with a defined distribution of wavelengths that yields a particular white point. It is intended to be close to (European) full-sun daylight.

I find that if I photograph a Colorchecker, full sun in (New England, North American) daylight, and use Lightroom's white balance tool (on the fourth white patch for the CC) that I get a temperature of 5400 or 5500. Can anyone illuminate this discrepancy?

Many thanks,

Michael Morrsion

[sandymc]

I may be misinterpreting what you're actually asking, but the simple answer is that Adobe's definition of white as it applies to a white balance tool is 5500K. So far as I'm aware the LR white balance tool is no more complex than that.

Sandy

[digitaldog]

Hello,

As I understand it, D65 is a spectral curve with a defined distribution of wavelengths that yields a particular white point. It is intended to be close to (European) full-sun daylight.

I'd have to do some searching on Google, but I recall that the CIE did actual measurements in a number of areas to produce their specifications for D65 (for that matter all standard illuminants). That said, I don't expect you'll find an exact match anywhere else, considering there are a number of environmental conditions that affect the color of white here. You should be "close enough". Also, with respect to Adobe or anyone else's products, again, YMMV in terms of what the values provided say as they correlated to but are not going to exactly be the specification of the CIE (and in Adobe products, the values are provided not as standard illuminants but as correlated color temperature using Kelvin).

[mcmorrison]

Hello Sandymc and Andrew,

Thanks for your thoughts. I too have wondered whether the temperature numbers in LR (or ACR) are close to real world numbers, but they always seem pretty close when I white balance on a Solux or other tungsten. My thought at the moment is that D65 is daylight at the blue end, such as might be seen at high altitude, etc. D50 is closer to daylight at sea level (at least here in New England, where I get around 5,500). The particular daylight condition D65 is meant to reflect remains mysterious for now. . .

Best,

Michael

[JeffKohn]

I've often wondered about this discrpency as well. Every camera and RAW converter I've ever used had neutral daylight at around 5500 or so. 6500 is what I would associate with slightly overcast light.

On the other hand, televisions and computer displays look pretty warm when calibrated to D65, I would never calibrate them to 5500 as that would look quite yellow. Makes me wonder if theres something about the way we perceive reflected versus emitted light.

[MPatek]

Just to add to what Andrew already said. CIE D illuminants are only mathematical simulations of various phases of natural daylight. Daylight illuminants (D-) are based on over 600 spectral power distributions measured (and averaged) at different locations and under various combinations of irradiation from sun and sky.
Since a simple correlation was found between the correlated color temperature (empirical extrapolation to a black body radiator) of daylight and its relative (measured) spectral power distribution, these theoretical spectral plots have been recognized as standards by CIE. As a result of being averages of various light sources, there may or may not be an actual light source that can reproduce them. Illuminants are defined by their specific spectral data and may not be physically realizable. If physical light source exists it is an source (e.g., illuminant A). In short, measured color temperature values will often not match with the standard daylight illuminant.
Regarding the white points in color managed applications, they should reflect the absolute color correlated temperature which is based on "color" of white (gray) patches. The main difference from illuminant values is that monitors are real radiation sources and thus what you what you measure or see in your monitor may not match with standard values. You should still be quite close though. Appearance of D65 source is closer to bluish, D50 to yellowish.

For a reference, here in Arizona (9:00 a.m., sunny, alt 2,400 ft.) I measured directly color correlated temperature of 5,700 K off the skies (EyeOne Pro spectrophotometer). This value will vary depending on geographic location, altitude, weather, time of the day.

For more details, see Color and Colorimetry

Marcel

_____________________________

Digital Photography Marcel Patek
http://www.marcelpatek.com

[mcmorrison]

Hello Marcel,

Thank you for your valuable post, and excellent website. I found your site quite valuable and illuminating—it answered my questions and more!

Best,

Michael

[mcmorrison]

Hello,

After reading Marcel's excellent site, I have learned to distinguish a few terms: sunlight, daylight, and skylight. Sunlight is direct sun, and it runs near 5500. Skylight is light that comes from the sky, and has no direct sun, i.e. in the shade facing north. Daylight is a combination of sunlight and skylight. Skylight is the bluest, and closer to 6500. Daylight is in between. D-65 is a spectral distribution curve constructed from averaging many types of light, and thus may not have any real analog. But if we are looking for something nearer to D-65, we might use skylight, rather than sunlight or daylight.

Here in Maine, we have had close to 2 weeks of solid rain, with another week predicted. As a result, I have not had a chance to test this out just yet.

Best,

Michael

[digitaldog]

Sunlight is direct sun, and it runs near 5500. Skylight is light that comes from the sky, and has no direct sun, i.e. in the shade facing north. Daylight is a combination of sunlight and skylight. Skylight is the bluest, and closer to 6500. Daylight is in between. D-65 is a spectral distribution curve constructed from averaging many types of light, and thus may not have any real analog.

Also keep in mind, any color you define using CCT (Kelvin) is a range of colors. Any color defined using a standard illuminant (an SPD) is an exact color. 5000K is a range of colors. D50 is an exact definition of a color. So this can get messy.

[Ethan_Hansen]

Andrew makes a good point regarding CCT values. Just because a light source (or RGB value calculated by a RAW processor for that matter) claims to be 5000K or any other color value tells you next to nothing about the spectral makeup of the light. A light may have a blackbody spectral output (e.g. sunlight, tungsten, a campfire), or it may contain a small number of spectral spikes the sum of which create the listed color temperature (fluorescent bulbs), or somewhere in between such as the multi-phosphor fluorescent bulbs used in high quality viewing booths.

This becomes important when dealing with metameric materials. Viewing a print under spectrally spiky lighting can give unpredictable results. Your 5000K viewing environment may well render print colors differently than others. The Color Rendering Index (CRI) of the bulbs used is a rough guide, but not something to rely on.

A case in point are the venerable Kodak 18% Gray cards. View them  under filtered tungsten lights or with really good fluorescents and the cards are dead neutral. Use a lesser quality fluorescent tube and the Kodak cards take on a green cast. You can use this to your advantage as a quick, accurate, and instrument-free method for light quality of viewing areas. I carry a small Kodak card in my bag when visiting client locations for the first time. A quick check in their viewing booth(s) goes a long way towards debugging any print color matching issues.

[mcmorrison]

Hello,

Yes, the difference between a CCT and, say, D-65 is becoming clearer.  One of the concepts that is emerging is that the "Planckian Locus" is the set of distributions that would be emitted by a true blackbody over a range of temperatures. As I now understand it, this locus is also what we visually experience as "white". Move off the locus and the white starts to appear to have a green or magenta cast. Hence the two WB sliders: one for finding the CCT, and one for finding the Planckian locus at that CCT. (Spikey light is approximate at best with plenty of caveats). As Andrew says, D-65, or D-50 are specific points, not exactly on the locus,  but close. As I understand it, D-65 is just a tad green relative to the locus.

Thanks again,

Michael

[madmanchan]

The CIE D-series of illuminants are standards. Tables of numbers. In general, real daylight will not match any of the D illuminants, but often will come close. As noted above, it depends on several factors, time of day, atmospheric conditions, location, angle, field of view, etc. There are infinitely many lights that have correlated color temperature values of 5500 K, including some old fluorescent bulbs I have sitting around here that make photographic prints look rather nasty. On a "normal nice" day of mixed sun and clouds you would expect a CCT in the range of 5400 to 6000 K. As it gets cloudier you get more "blue-ish" light and the CCT goes up.

[bjanes]

The CIE D-series of illuminants are standards. Tables of numbers. In general, real daylight will not match any of the D illuminants, but often will come close. As noted above, it depends on several factors, time of day, atmospheric conditions, location, angle, field of view, etc. There are infinitely many lights that have correlated color temperature values of 5500 K, including some old fluorescent bulbs I have sitting around here that make photographic prints look rather nasty. On a "normal nice" day of mixed sun and clouds you would expect a CCT in the range of 5400 to 6000 K. As it gets cloudier you get more "blue-ish" light and the CCT goes up.

To add to the above discussion, see this link, which contains an illustration from Wyszecki & Stiles, 1982, a standard reference.

D65 approximates noon daylight
D55 approximates noon sunlight

Daylight consists of sunlight plus skylight. If you want to expose with sunlight, you could use a shift of sunlight coming into a room through a window or skylight, which would obstruct most of the skylight. To expose with skylight, you could use light from a north facing window. Daylight may to modified by surrounding foliage or buildings.

Bill