• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

About "white balance" tools

Doug Kerr

Well-known member
Prompted by some recent forum traffic regarding "white balance tools". I thought I would give a little review of the matter of "white balance color correction" and some of the tools used to make measurements needed to pursue that process. I will of course begin with some background, as is my wont. And of course I will greatly "simplify" some complicated technical notions.

BACKGROUND

Chromaticity

I will speak often here about chromaticity. Chromaticity is one aspect of the color of light, the aspect that "distinguishes red light from blue light" and that distinguishes "red light from pink light". It is the property of light that lay people think is meant by "color"; we of course recognize that color also includes the separate aspect of luminance.

We note that chromaticity is a subjective property, defined only in terms of human visual response. The chromaticity of an instance of light is not a simple physical property (as would be, for example, the frequency of an electrical sine wave).

But the chromaticity of an instance of light is determined, in a complicated way, by the spectrum of the light (the distribution of the power in the light with wavelength).

Reflected light

The spectrum of the light reflected by an object (and thus the chromaticity of that light) is directly determined by the spectrum of the illuminating light together with the "reflective spectrum" of the surface.

Can we then say that the chromatic of the reflected light is determined by the chromaticity of the illuminating light together with the "reflective chromaticity" of the surface? No, it doesn't work that way - at least not "exactly". But in many cases it works "sort of" that way, and we will here deal with that situation as if it were "exact".

Chromatic adaptation

In any case, we can readily see that the chromaticity of the light reflected from, for example, a sheet of "white" paper on our breakfast table illuminated by mid-day sun through the window will be different from the chromaticity of the light reflected from that same sheet of paper when it is, in the evening, illuminated by light from an overhead incandescent lamp.

Yet most likely in those two situations a human viewer will recognize the paper as "white". How is this possible?

The answer is in a wondrous property of human perception called chromatic adaptation. To greatly simplify a very complicated topic, the human visual system "deduces" the chromaticity of the illuminating light (by noting the chromaticity of the light reflected from familiar objects in the larger "scene") and then in effect concludes what the "reflective chromaticity" of the paper must be.

But in a photograph

Now suppose we take a photograph of the breakfast table, including the famous sheet of paper, under the overhead incandescent light, make a print, and send it to a friend.

The friend views it on her breakfast table, illuminated by the late morning sunlight coming in through the window. Her visual system cannot "deduce" the chromaticity of the light that illuminated the paper when the shot was taken by observing surrounding items - they are not included in the photo. Rather, her visual system deduces the chromaticity of the ambient light at her viewing location (where there are many objects within view).

The result is that the paper looks to the viewer reddish-yellowish, not "white". And in general, that is not what we want.

White balance color correction

Here is how we overcome that presumably-unwanted effect. When we are getting ready to take the shot, we measure the chromaticity of the illuminating light. We then (in the camera, or in post-processing software), in effect deduce the reflective color of each area in the photo (conceptually "backing out" from the recorded chromaticity of the reflected light the measured chromaticity of the illuminating light), and then modify the chromaticity recorded for that area (from the light captured by the camera) to be that which would seem to the viewer to indicate the real reflective color of that area, when the "print" is viewed under some assumed ambient chromaticity.

Ah, but there's the rub. What do we assume about that chromaticity? We do not know if our viewer will view the print in an area generally illuminated by late morning sunlight (as we assumed earlier) or at the breakfast table late in the evening, under the overhead incandescent light.

So, under the valuable rubric of "better the devil we know than the devil we don't", we do this based on the assumption that the image will be viewed in a context where the ambient illumination has the chromaticity of the white point of the color space in which the image is delivered.

MEASURING THE CHROMATICITY OF THE ILLUMINATING LIGHT

So, how can we measure the chromaticity of the illuminating light on the scene being photographed? Well, there are several possible ways.

A colorimeter

We can use a colorimeter, whose receptor we expose to the light incident on the scene. It reports, in a certain scheme of notation, the chromaticity of that light.

Before we go on, let us note this important consideration. If in fact the light incident on the scene comes from different sources (from different directions), with different chromaticities, we want our (single) measurement to represent the joint impact, chromaticity-wise, of those multiple sources on the chromaticity of the light reflected from the various objects. If those objects have the classical diffuse reflecting property, then the impact of the various sources is proportional not only to their "potency" but also to their angle of incidence onto the surface.

Our colorimeter can take all that into account if the response of its receptor varies with the angle of arrival of multiple light components according to the cosine of their angles of arrival. So colorimeters are made to do that. Hold that thought.

In any case, there are at least two practical problems with this approach:

• Colorimeters are rather expensive.

• It is in general not easy to take a chromaticity determination made by a colorimeter and explain it to our camera, or our post-processing software

But we can solve both of those problems by using the next method.

The camera as colorimeter

But we can turn our camera into a colorimeter. Suppose we put in front of the camera lens an optical unit arranged to collect the light that falls upon it, "weighting" each ingredient proportional to the cosine of the angle from which it arrives, mix all the collected light together, and present it to the lens as a uniformly-illuminated disk. (It will be out-of-focus, but that doesn't matter. It will still "fill the frame" with its chromaticity.)

This optical accessory is often called a "white balance diffuser".

We then place the camera, with the white balance diffuser in place, at the subject location in the scene, with the diffuser oriented to receive the illuminating light much as the pertinent parts of the subject will. (Most often, an appropriate orientation is to have the diffuser facing the spot where the camera will be for the actual shot.)

Modern cameras are arranged to take their observation of the chromaticity they observe in that situation (perhaps by way of a frame captured in that situation) and (in one of several ways) introduce it into the in-camera white balance color correction process.

Or, we can take the frame that captures that chromaticity and use it in our post-processing software to guide it in performing the white-balance color correction process.

White balance diffusers are available from various manufacturers. One unit that has a long reputation of careful manufacturer and good performance is the ExpoDisk diffuser, made by Expoimaging.

A more indirect approach

Another approach is to include in either the actual scene to be shot or (more commonly) in a preliminary shot an object that is "spectrally neutral". That is, its reflectance is the same for each wavelength over the visual wavelength range.

The consequence is that the chromaticity of the light reflected from it is the same as the chromaticity of the light that illuminates it.

For various reasons (which are beyond the scope of this note - some are folkloric), it is the custom to use for this purpose an object that not only is spectrally neutral but which also has an overall reflectance substantially less than 1.00 (100%). Accordingly, we see such a "target" as being "gray", and these are therefore commonly called "gray cards".

There are at least two ways to utilize a neutral target for determination of the chromaticity of the illumination on the scene being/to be shot.

• We can have the target illuminated by the illuminating light and then take a shot of it with our camera, having it fill at least the part of the frame the camera will use from a "white balance test shot" to determine the chromaticity of interest. It is not necessary (nor actually even desirable) for the target to be in focus when we do this. The image we capture can be used to guide the camera in doing the white balance color correction, or can be made available to our postprocessing software to the same end.

• We can place the target in the actual scene, with the illumination in place, and then take a shot. We can then remove it and take the real shot. The, during post-processing, we will load the preliminary shot and use the "eyedropper" tool to note the recorded chromatic of the neutral target, using that to guide the application of white balance color correction to the actual shot.

Neutral targets ("gray cards") are available from various manufacturers. One series with a long history of careful manufacture and good performance is the WhiBal test target series, made by Michael Tapes Design (yes, our colleague Michael Tapes).

Measurement from the camera position with a diffuser

From time to time, various manufacturers have presented camera attachments (in the general vein of a white balance diffuser) which purportedly will allow us to, with our camera, located where it will be for the actual shot, determine the chromaticity of the illuminating light onto the scene. How convenient.

But in fact, this in general cannot work. Much of the light captured by the device and presented to the lens is light reflected from the scene, and of course its various components will have chromaticities resulting both from the chromaticity of the illuminating light and the spectral reflectance (we might imprecisely say "reflective chromaticity") of the various objects in the scene. Thus the chromaticity of the light gathered by the device cannot tell us what we need to know: the chromaticity of the light illuminating the scene.

Yet, in many cases, such an approach turns out to give a chromaticity determination that is "not bad". How can this be?

Well, in many cases (typically in outdoor photography), the light incident on the camera position will be very much like the light incident on the subjects (whose chromatically we want to determine).

And, although the manufacturers of these devices often emphasize their "concentration" on the "light from the subjects", they in fact often have very broad acceptance angles.

Thus, a measurement from the camera position with such a device on the camera may mostly capture the illuminating light incident on the camera location. And if that light has the same chromaticity as the light illuminating the subjects, we get a quite usable result.

But that good result is contingent on circumstances that cannot be depended on over a range of photographic tasks. And thus the notion of a camera accessory that will reliably allow the chromaticity of the light illuminating the subjects do be determined from the camera position is, overall, just wishful thinking.


Best regards,

Doug
 

Jerome Marot

Well-known member
But in a photograph

Now suppose we take a photograph of the breakfast table, including the famous sheet of paper, under the overhead incandescent light, make a print, and send it to a friend.

The friend views it on her breakfast table, illuminated by the late morning sunlight coming in through the window. Her visual system cannot "deduce" the chromaticity of the light that illuminated the paper when the shot was taken by observing surrounding items - they are not included in the photo. Rather, her visual system deduces the chromaticity of the ambient light at her viewing location (where there are many objects within view).

The result is that the paper looks to the viewer reddish-yellowish, not "white". And in general, that is not what we want.

A print taken under candlelight appears too orange when viewed under daylight. But then, why doesn't a print taken in daylight appear too blue when viewed under a candle?
 

Doug Kerr

Well-known member
Hi, Jérôme,

A print taken under candlelight appears too orange when viewed under daylight. But then, why doesn't a print taken in daylight appear too blue when viewed under a candle?

Well, if that's actually so, I'll have to ponder why!

It may be that the viewing environment is not "candle-lit", just the illumination of the print. It is after all the viewer's assessment of the ambient chromaticity in the "viewing room" that is at work here.

But a very interesting question.

Best regards,

Doug
 
Top