# Is it possible to determine the color temperature of a light source from a photo?

I have several recessed LED lamps that I'd like to assess for color temperature so I can purchase replacements. The lights are advertised as having a "color temperature" of 2700K and 3000K among others (example here). Is there a way to detect the color temperature from a photo or the lamp, perhaps by taking a photo side-by-side with a grey card?

• I'm voting to close this question as off-topic because this is about using cameras as color-temperature measuring devices. Sep 30, 2019 at 2:56
• The temperature should be printed somewhere on the bulb itself. Sep 30, 2019 at 3:34
• @mattdm Determining color temperature of light sources is as basic a task for producing photographs as metering a scene for brightness is. Digital cameras do it internally every time a photo is taken. As such, one would think that using a camera to determine the color of a light source is not off topic for this forum. Sep 30, 2019 at 20:44
• Possible duplicate of Is there a way to figure out a lamp's color temperature from a photo? Sep 30, 2019 at 20:45
• Cameras don't need to determine accurate color temperatures as long as the photos look okay. Using three different lenses with the same lamp and white card, my camera reported temperatures of 2805K, 3164K, and 6490K after setting custom white balance with each lens. These values are all over 100K from the bulb rating. Oct 1, 2019 at 2:02

Two possibilities:

1. You camera has a "Custom white balance" and will display the corresponding color temperature(*): take a picture of a gray chart or equivalent, apply custom white balance, check the result temperature.

2. You can set arbitrary color temperatures in the camera's WB: shoot a series of pictures of a grey chart over a range of temperature settings, check which picture is the most neutral (R=G=B).

## Edit: testing method #2

It turns out that my camera will display RGB histograms, so there is no need to transfer the pictures to a PC to check things out, everything can be done with the camera display.

So, a quick test with my bed lamp, shooting my color reference, changing the color temperature (which is conveniently add to the display):

The green histogram doesn't move, but the red and blue one eventually get aligned at 2700K.

Coincidence? This is exactly what the lamp markings say:

## Edit#2: further tests

Testing with:

• two lenses:
• Canon 35mm IS STM (the one used above)
• Sigma 17-70mm
• two fluorescent lamps:
• the 2700K one used above
• a "Cold daylight" one (no markings on the lamp, but rated 6700K on the box).

Findings:

• Both lenses produce the same TC readings
• Accuracy at the other end of the spectrum is worse, the camera has to be set to 7800K to align red and blue histograms.
• At the high end of the spectrum (8000K) a wide variation of the TC settings results in minute histogram shifts, while at the low end (2700K) a single step (100K) has a very visible effect, so the inaccuracy at the higher temperatures would not be that much of a problem, lamps with a 500k difference wouldn't make that much difference in the rendered colors.

(*) Not my Canon 70D, unfortunately

• I'd say the easiest approach would be to take a RAW picture and do post-processing to determine the color temperature. Sep 30, 2019 at 11:24
• @Juhist Easiest is definitely #1 if camera allows. Your solution can be easier than #2. Sep 30, 2019 at 11:48
• @xenoid Notice that when Red and Blue are roughly equal, green is still slightly stronger. This indicates that your light source does not fall directly along the color temperature axis, but also has a bit of green tint in its output along the Green ←→ Magenta axis that is roughly orthogonal to the Blue ←→ Amber color temperature axis. Sep 30, 2019 at 20:39
• You would need to apply one or two ticks of WB correction towards magenta to equalize the green level with the Red and Blue levels. Sep 30, 2019 at 20:46
• @xenoid Did you know the lamp temp before performing your experiment? If so, results may be biased. Why did you stop after getting the desired results rather than continue with 2600K and 2500K? Have you tried using different lenses to see if they all produce the same results? Maybe the unnamed lens you used this time is the only one that can reliably measure color temperature. What about different temperature bulbs? Maybe your camera is accurate only around 2700K. Oct 1, 2019 at 20:53

In order to fully compare the light output of two sources, one must consider both the temperature (Blue ←→ Amber) and tint (Green ←→ Magenta) of the light a source emits.

Measuring color temperature alone will not tell you all of the differences between two light sources. Color temperature is but one axis in color space that includes the colors emitted by black body radiators when heated to specific temperatures expressed in degrees Kelvin. The glowing gases on the surfaces of stars, for example, are black body radiators. So are most metals when heated until they begin to glow, then melt, and finally turn to vapor if heated hot enough.

Most natural light sources emit light only along this Blue ←→ Amber axis. But other light sources can also have a strong "tint" along the Green ←→ Magenta axis in the light they emit. For example, the LED lighting currently used for stage illumination in a lot of small night clubs can have a much more magenta tint to it than a black body radiator will emit at any temperature. Typical old style fluorescent lights, on the other hand, emit a much greener tint than a black body will radiate. This "tint" axis is more or less orthogonal to the "color temperature axis/curve in color space charts.

The longer curve is a representation of the color temperature axis. The shorter hash marks that cross it are oriented in the direction of the tint axis at different points along the CT axis.

So how do I measure the total white balance of a light source using a camera?

One way to do it is to take a photo of a color neutral object such as a "gray card" and save the raw data collected by the sensor. Then open the raw file in a raw conversion application, such as Lightroom, and adjust the color temperature and white balance correction values until the RGB values, represented by numbers such as (228,247,218) are all equal (233,233,233). The color temperature of the light source is whatever CT value, expressed in degrees Kelvin, you are using on the CT adjustment. The tint of the light source is indicated by the amount of Green ←→ Magenta correction applied. With Canon's Digital Photo Professional, which is what I often use to process raw files, each unit of Green ←→ Magenta correction represents a five Mired color correction filter.

Another method would be to use the EXIF info included in the 'Maker Notes' section of many cameras to see the Red, Green, and Blue measured values used by the camera to make a color neutral object appear color neutral in the resulting photograph. To do so take a photo of a color neutral object that fills the frame. Then set the camera's "Custom White Balance" option to use the photo of the color neutral object as a reference. Once you've done that, take another photo of the same color neutral object under the same lighting. The color values used to produce the second image can then be used to calculated the color temperature and tint of the light source, but that involves a lot of transformational math that is probably beyond the scope of an answer here.

Not very many applications that show EXIF data will display that information in the EXIF data, though. Some raw convertors will open raw files using that value in the EXIF info as the default value if the selected options for opening raw files specifies to use camera calculated WB. Among the tools I routinely use, it isn't displayed. But if I upload an image taken with my Canon EOS 5D Mark III to Flickr and allow the EXIF info to be displayed, anyone can click on the "Show EXIF" link and then scroll way down to see it.

Here's a piece of a screen shot from a portion near the end of the EXIF info for this image at flickr:

Notice that the "Color Temperature Auto" and "Color Temperature Measured" values are identical. That indicates that when the camera is set to "Auto WB" the measured values for Red, Green, and Blue are used to calculate the color temperature and tint that are then applied to the processed image (either the jpeg preview attached to the raw file, or a jpeg image).

• See, this is why I think it's better to just close the question. How will this help in determining which replacement bulbs to buy, when the bulbs will be listed with no more than a color temperature rating in kelvin? Oct 1, 2019 at 0:03
• @mattdm Some specialty bulbs are rated with more than just a CT rating. They are D50 compliant, or D65, or whatever. D50 is not the same as 5000K. D65 is not the same as 6500K. The spectral distribution must also be compliant with the standards, and the peak white point is not exactly along the CT axis, it's slightly greener. Oct 1, 2019 at 4:30
• Sure, but what are the odds that that's relevant to the actual question here? Oct 3, 2019 at 2:07