Empirically judged, I would assume that the human vision can't compensate the color of 1850K (candle light) equally efficient as the color of 3000K (incandescent lamp). So, is there a range in which white balance can be practically applied?

In other words: outside which Kelvin range a possible white balance compensation would produce an unrealistic color cast, rather than expected compensation towards "white"?


3 Answers 3


This is described by the Kruithof curve, and the answer depends on the brightness as well as color temperature. From Wikipedia:

public domain image

At the brightness of direct sunlight (along the top of the chart), anything from 4500K up will appear basically natural. At low light levels, though, much warmer (in the art sense — lower K!) light feels more natural.


This one is complicated. The natural response of your eye in well-lit conditions is known as the photopic response.

Image courtesy wikipedia

When the light goes down a bit, your eye switches to Mesopic vision, essentially generating and using a good number of rods to supplement the cones. Your cones have very good color sensitivity, but are inefficient and bad in the dark. Rods have bad color, but great sensitivity. Mesopic vision "lasts" until about the light level of a full moon.

After mesopic vision comes Scotopic vision which is purely from your rods. Colors desaturate and you loose red response almost completely, with a peak sensitivity at a teal color.

Image courtest wikipedia

A color temperature greater than approximately 7000K-8000K is cooler than your eye is used to seeing and will appear unnaturally blue. In dark conditions you may not be able to distinguish it well from a temperature of 5500K (daylight). Camera noise and a loss of color saturation in the dark is also an important factor. A temperature lower than about 2500K is mostly producing energy in the infrared. Such an object would appear strikingly red, perhaps unnaturally so.

Many objects, sources, etc, do not produce black body curves. CFLs and LEDs typically do not, while incandescent bulbs produce somewhat modified ones.

I hope this sufficiently answers your question.

  • \$\begingroup\$ Thank you for the physics behind that topic! So, 2500K...8000K then. But, as far as I understand, you are answering more about visible colors in general, not more specific about colors which can be perceived as white by human vision. \$\endgroup\$ Commented Jul 2, 2015 at 17:10
  • 1
    \$\begingroup\$ Colors which can be perceived as white is very tricky and not my area of expertise - it is more medical optics where I study the optics of lenses. As far as I understand your brain figures out the final color of things based on past experience, expectations, etc. For example if you see an egg, your brain can correct that to white because it knows what an egg is. If you saw a mythical dragon your brain would tend towards the "raw color response." \$\endgroup\$ Commented Jul 2, 2015 at 17:36

There is no easy answer. If you were viewing by unnaturally blue light, you might get used to it and everything seems normal after a while. but, you could not tell the difference between a blue object and a grey object, or differences in blue hue between them. Now consider that other colors have some blue component... the information is lacking, and your ability to judfe correct color depends on your knowledge of the objects. If all seems OK, you'll be surprised when confronted with an unknown object when you suddenly can't tell. Or, fooled by an object that is the wrong color.

  • \$\begingroup\$ sorry, but this answer sounds like "i don't know" \$\endgroup\$
    – szulat
    Commented Oct 28, 2016 at 17:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.