8
\$\begingroup\$

I can set my white balance camera dial to sunlight, to overcast sky or to a tungsten lamp ensuring there's no bluish or yellowish tint and the white is really white. I know I can fine-tune these values further when developing from RAW.
All these light sources have continuous spectra (unlike fluorescent or HID lamps which is another story).

But do I get identical colours (namely blue) when taking the same scene under an overcast sky, under direct sunlight (white balanced to about 5600K) or with a halogen lamp (white balanced to 3000K)?

Note: I'm asking because there are different spectral compositions and I don't know if camera can effectively compensate these variances.

\$\endgroup\$

3 Answers 3

9
\$\begingroup\$

No. You won't get the exact same colors with white balance only.


1 - Built in settings

The built-in white balance settings are a very, very generic correction. If you are in a hurry, well, they are better than nothing.

The first basic color correction these settings make is color temperature. But not always the problem is the color temperature, but also, for example, the tint.

2 - Custom White balance

You can make a custom white balance with a grey card to have better specific results.

A white balance is a simple correction between the 3 RGB chanels. Let us imagine that corrects an image using simple curves:

(These graphs are just to get the idea)

enter image description here

Here is an explanation of setting a custom white balance: Color issue: studio images have a pink hue

3 - Profile

But to achieve more controlled color in different light situations you need a profile. A profile makes more elaborated corrections in different zones. Imagine that the camera accurately captures bright blue tones, but darker ones need corrections.

enter image description here

So you need a color chart, not just a grey card. One standard is Macbeth. These charts need to be printed with high-quality pigments, so each patch has a standardized value, that the software can use to adjust curves on the color profile.

enter image description here

The software analyzes the target and makes a profile because it knows what the color should be.

A good color chart is the X-Rite Color Checker Passport. It includes software to make a profile for the combination of your camera + the specific lighting situation in which you take the photo. The color changes can be different from flash to flash, from different brands, different models, or even if one flash is old and the other is new.

4 - Extreme light

But even in some extreme light situations, you can not achieve the same colors, for example, I would not use sodium street lamps for a product shoot. Some types of lighting lack some wavelengths, so some physical colors can not be reproduced.

There are some standards to know this quality of the illuminants, for example, the CRI value. Color Rendering Index.

5 - Dynamic Range

You also need to take into account how much your camera can really perceive inside one single shoot.

That is the dynamic range. And depending on the light situation you have and the settings on your camera you can crop a part of the colors.

This depends mainly on the correct exposure. You can have a good white balance, but if the exposure is not right...

6 - Your camera itself

The sensor and the way your camera processes an image vary, from brand to brand, model to model.

If you add some personalized settings like "portrait", "landscape", "neutral" you can have slightly different colors. Some settings can affect just the jpg or flavors of a RAW file.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ Many cameras have a tint adjustment as well as color temperature. All of my Canon bodies can do +/- 45 mireds in 5 mired increments on the magenta←→green axis. \$\endgroup\$
    – Michael C
    Commented 2 days ago
  • 1
    \$\begingroup\$ Some of the presets also seem to have minor amounts of M←→G adjustments built in to them, particularly 'White Fluorescent' and 'Cloudy'. \$\endgroup\$
    – Michael C
    Commented 2 days ago
3
\$\begingroup\$

In these three circumstances, nothing can get you identical colors all over the image because the lighting setup is totally different:

  • overcast sky - one huge softbox, about 6500K
  • direct sunlight - one bare light, about 5500K-6000K, and one huge softbox, about 15000K...27000K (clear sky) or 6500K (overcast)
  • halogen light - one bare light, 3000K

White balance, curves and other global color corrections can only account for the different color temperature of one light source. They cannot fix differences caused by different angle or nature of shadows.

In case of direct sunlight, the presence of a secondary light source acting as fill light with different color temperature (albeit a few stops weaker than main light) is going to skew hue in shadow areas to be different from sunlit areas. The only cases free of this phenomenon are completely flat subjects, so there are no shadow areas, or subjects completely in shade, so the direct sunlight is actually eliminated.

In the other two cases, a bare halogen light renders shadows different (harsh) from overcast (soft, barely distinguishable). These two can be likened by either diffusing the halogen light (with a diffuser much larger than the subject) or the other way around, restricting the ambient light to fall from only a narrow angle.

\$\endgroup\$
2
\$\begingroup\$

The short answer to your question in "No". White balance can not compensate for color differences due to different light sources.

The slightly longer answer is that white balance is a set of relative exposure gain factors applied to the image post-exposure.

The spectral radiance of the scene to be captured is the integral of the spectral power distribution of the light source used to illuminate the scene and the spectral reflectance of the objects in the scene. The Spectral power distribution of the source effectively defines the color of the scene at the focal plane. If the source changes, the relative radiance of objects in the scene will change compared to each other creating a multidimensional problem.

In theory your direct sun vs. shade example would be able to be compensated for since the source is really the same in both pictures (daylight) and is comprised of the same eigenvectors allowing for a linear solution such as a matrix or even WB gain factors. In practice it would be very difficult to match the image colors perfectly because reflections from near objects (grass, trees, a painted wall of the building casting the shade, etc.) will effectively modulate the shade such that it's not really a pure daylight spectra anymore.

\$\endgroup\$
5
  • \$\begingroup\$ Daylight != skylight. \$\endgroup\$ Commented Mar 4, 2017 at 15:05
  • \$\begingroup\$ @EuriPinhollow What does that even mean? Are you advocating using a "skylight" filter with a digital camera? Or something else? \$\endgroup\$
    – Michael C
    Commented 2 days ago
  • \$\begingroup\$ @MichaelC "your direct sun vs. shade example", "the source is really the same in both pictures (daylight)". Literally as I said sky-diffused sunlight is never same as the sunlight, particulary when it's overcast sky. \$\endgroup\$ Commented 2 days ago
  • \$\begingroup\$ That's like saying sunlight on the moon is the same light source as sunlight on the Earth's surface - when it isn't. The spectral distribution is different due to the filtering of the Earth's atmosphere. Any change in atmospheric conditions also changes the spectral distribution of the light that makes it through the filtering elements in the atmosphere. \$\endgroup\$
    – Michael C
    Commented yesterday
  • \$\begingroup\$ Wait. Are you saying != is the same thing as ≠? And it's not my example. I didn't write the answer above. \$\endgroup\$
    – Michael C
    Commented yesterday

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.