White Hot: The Accuracy of Color Temperature

Mick West

Administrator
Staff member

Source: https://www.youtube.com/watch?v=VzcodRCLOE8


Something that occasionally comes up in various conspiracy theories is an attempt to determine temperature from the color something appears in an image or video. This is based on very reasonable science in that the color of light that something like a piece of steel (or really any metal) emits has a very specific correlation to its temperature.

However, as I demonstrate in the above video, this only works well to the naked eye and tends to go wrong in videos or images unless a lot of care is used in the exposure. Since cameras (and cameramen) generally set the exposure for the overall scene then this leads to smaller hot regions of the image being greatly overexposed. Hence they appear a lot hotter than they actually are.

Metabunk 2018-12-07 13-16-46.jpg

Here's a more focused version of the above video, that contains a direct comparison

Source: https://www.youtube.com/watch?v=zHWMYtdTLyc


Some key temperature:

Metabunk 2018-12-07 08-36-40.jpg
That's nearly 1200°C, but on the iPhone it looks like a molten blob over 1300°C

Metabunk 2018-12-07 08-37-50.jpg
At 1100°C the better exposure is a bright orange, almost yellow. But the iPhone still shows incandescent molten temperatures.


Metabunk 2018-12-07 08-40-44.jpg
At 1000°C we are correctly orange, but the iPhone is showing nearly 300° hotter!

Metabunk 2018-12-07 08-43-26.jpg
950°C, the iPhone still thinks it's over 1100°C

Metabunk 2018-12-07 08-44-11.jpg
At 900°C things start to look better

Metabunk 2018-12-07 08-45-02.jpg
And at 800°C the colors are similar
 
Last edited:
Also, it is worth making clear that, even if you can accurately assess the color of a glowing object, you still cannot accurately determine its temperature unless you know its exact composition because an object's glowing color is the result of how much energy such object is emitting and an object's emissivity depends on its composition (e.g., steel has a different emissivity than aluminum). Thus assertions by conspiracy theorists of the temperatures of glowing objects and substances based on photos and videos are doubly flawed in cases where the composition of such objects are not precisely known (as is the case with WTC debris).
 
Also, it is worth making clear that, even if you can accurately assess the color of a glowing object, you still cannot accurately determine its temperature unless you know its exact composition because an object's glowing color is the result of how much energy such object is emitting and an object's emissivity depends on its composition (e.g., steel has a different emissivity than aluminum). Thus assertions by conspiracy theorists of the temperatures of glowing objects and substances based on photos and videos are doubly flawed in cases where the composition of such objects are not precisely known (as is the case with WTC debris).

Really? Emissivity would seem to only alter the brightness, not the color.

And what does this mean on a practical basis though? What materials might be confused? Does aluminum at 600°C actually look different to steel at 600°C?

Hmm, tricky to test, as I'd have to heat it in the dark to even see it....
 
Really? Emissivity would seem to only alter the brightness, not the color.

And what does this mean on a practical basis though? What materials might be confused? Does aluminum at 600°C actually look different to steel at 600°C?

Hmm, tricky to test, as I'd have to heat it in the dark to even see it....

Metabunk 2018-12-07 12-05-00.jpg

Result. Aluminum on top, steel on the bottom. The emitted light looks like the same color to me.
 
Really? Emissivity would seem to only alter the brightness, not the color.

And what does this mean on a practical basis though? What materials might be confused? Does aluminum at 600°C actually look different to steel at 600°C?

Hmm, tricky to test, as I'd have to heat it in the dark to even see it....

I think you are generally right and I was mistaken in how I described the relationship in my post. As I understand it after reading more deeply, almost all objects will emit black body radiation at the Draper Point (798 K) and thereafter follow the same temperature dependent color curve. To the naked eye, there can be some exceptions to this as certain objects are not emissive enough to visibly glow red at the Draper point.
 
Last edited:
I think what you are referring you is 'intensity', @benthamitemetric

To my understanding, and please correct me if wrong, but intensity in physical terms is the density of the radiating energy (the energy flux). In common language we call that "brightness" when talking about light.

More brightness does not affect the wavelength, ie. the colour of the light because more brightness/intensity simply means the amount of photons flowing through a given area is increasing, and we perceive that increase as "brighter". But that does not affect wavelength.

Just as more photons hit our eyes, a sensor of a camera will also register more photons as being brighter. But again, it will not affect the wavelength and the colour of the light.

For example, if the intensity is very low, there won't be enough photons for human eyes to register, even if individual photon wavelengths are very short and thus high in energy. It's similar to how extremely hot gas molecules won't produce any sensation of heat if the density of the gas is extremely low.

'Emissivity' however is merely the ratio of how well a black body is able to release thermal energy in accordance to the Stefan-Boltzmann law. 0 means no energy, and 1 means maximum possible amount for the given temperature.

What we perceive as white can come from two things:
• From a wavelength perspective it's the "sweet spot" where the red, green and blue wavelengths are in balance.
• Sensory overload, or overexposure. In such case the wavelengths does not need to be in balance as long as all wavelengths comes in intensity greater than the eye or exposure settings in a camera can handle.

Thus, estimating temperature by colour from photos can be tricky and misleading because:
• Cameras can have various exposure settings and limit in dynamic range (as shown in @Mick West's experiment with the two cameras).
• Cameras have white balance settings that will offset what the human eye consider to be white, essentially distorting all colours in the photos.

If you don't understand these things it can be confusing when watching illustrations as the one below:

Like, the reference chart used by Mick West in his experiment shows 1300°C (1 573k) being white, but in this chart between 5000-6000k is illustrated as white.

The difference as I understand it is that - intensity, the amount of photons, will overload the naked eye around 1300°C, thus making it appear white to us. But from a strict wavelength perspective, only somewhere between 5000-6000k (which happens to be the frequencies involved in normal daylight) will appear white to us because the red, green and blue light receptor cells in our eyes will be equally stimulated.

In other words, if you overexpose a photo of something glowing at much lower temperatures (letting more photons hit the sensor, increasing brightness), it can still appear white, and this needs to be considered when evaluating colour temperature from photos where no camera metadata is included.
 
Last edited:
The spectrum I used comes from the Wikipedia artlcle on "Red Heat" which is specifically for the color of iron. This spectrum is combining intensity (brightness) and color. An actual color temperature is just the color, but to the human eye (and the camera) anything actually hot enough to emit light above 1300° is going to be very bright, and hence white.

But things illuminated with that light are going to have a color cast. That's why you get "warm" light bulbs (which are actually the cooler temperatures).

Color is complicated.
 
Brightness affects color indirectly: The "true" color is the same for materials with different emissivities at the same temperature, but the way eyes, films or digital sensors record color is brightness-dependent.

All use 3 (or 4, rarely more) different light detectors, say, one for blue, one for green, one for red. However, the blue detector also gets triggered by green light, and vice versa, only less so. And, importantly, all detectors "max out" - they give their maximum response at some brightness, and if the light that hits them gets brighter, they can't return a larger signal and will record, incorrectly, the same brightness.

Now the thing is: Bright red light will get a much stronger response from the red sensor than the green sensor, but it will get some. As long as no sensor color is maxed out, the sensor composite stands a chance to record the correct color. However, once the brightness goes beyond the point where the red sensor maxes, out, the blue and green still have some capacity left: They'll return more and more - and that will affect the overall color perception. When, at long last, all sensor colors are maxed out, the system will return "white", even though the light is still firmly red, only very brightly so.
 
Back
Top