The ColorRight line of white balance tools - an update

[Doug Kerr]

When the ColorRight white balance tool first emerged (it was initially called the Color Parrot) I was fascinated - and baffled.

My understanding of white balance color correction (of the "theoretically ideal" form) was (and still is) this:

• In order to compensate for a difference in the chromaticity between (a) the illumination on the scene and (b) the chromaticity of the white point of the color space in which the image will be conveyed, we need to know the chromaticity of the illumination on the scene.

• That chromaticity can be measured by various techniques, which must be practiced at the subject location (since we are concerned with a property of the illumination upon the subject).

The manufacturer of the Color Parrot (fellow member Drew Strickland), however, emphasized a mode of use in which the device (essentially what we might characterize as a "white balance measurement diffuser) would be placed over the camera lens, with the camera placed and aimed as it would be for the actual shot, and a "measurement" frame taken.

I was unable to imagine how a measurement frame taken in this way would consistently tell us the chromaticity of the illumination on the subject.

So I raised the question of Drew of on what technical principles did this procedure depend.

No meaningful answer emerged, and none has emerged (to my knowledge) to this date.

I was, however, chided for attempting to apply scientific inquiry to the working of a device whose good performance was "well recognized" and which was being widely purchased by photographic enthusiasts, thereby spoiling the mellowness.

There was, by the way, no assertion that the working principles were "trade secrets" and thus could not be disclosed.

As I reflected on this conundrum (making numerous tests here), I concluded that the basis for the good color balance results widely reported with the use of this family of devices was most likely this:

In many situations, the incident light upon the subject and the incident light upon the camera position would be very similar. A measurement diffuser on a camera located at the camera position for the shot, and aimed in an arbitrary direction (including toward the subject), would essentially capture the illumination at that location (that is, on the camera). The chromaticity of that illumination would, in many such cases, be a useful approximation of the chromaticity of the illumination on the subject itself.

This premise has never (to my knowledge) been "confirmed" by ColorRight.

I noted in that regard that when the illumination on the scene is measured in the "traditional" way, it is desirable for the "antenna" of the instrument (such as a measurement diffuser placed on a camera) to respond to light arriving from different angles as would the subject itself. It is for this reason that "traditional" measurement diffusers have a "cosine" acceptance pattern - something that is usually achieved with some scheme such as an array of mini lenses on the face of the diffuser.

ColorRight, however, emphasized the fact that their device "concentrated its attention" on the subject. They never chose to describe exactly what technical property that was. I suspected it meant a relatively-narrow acceptance pattern. Tests done here with a sample unit (kindly provided by Drew) confirmed that.

No special step was taken in the unit (that I know of) to achieve that, nor would any be required. Rather, it is inherent in any basic diffuser (sheet of opal glass, for example). As I mentioned before, attaining a broader pattern (such as the "cosine" pattern) requires special features.

The basic units have most of their overall area masked, only a modest-diameter aperture being active. It has been intimated that this is in some way responsible for the "narrow" acceptance pattern, but it would not have that effect. (We have confirmed that with tests done here.)

Over the years, various versions of the ColorRight tools have been introduced. One includes a reflective target, which can be used for "gray card" measurement at the subject. In some eras, mention was made of the alternate possibility of using a camera equipped with the ColorRight device at the subject location (although its lack of a cosine acceptance pattern makes if far from ideal for this "ideal" technique). But of late, emphasis is placed on the "at the camera position" technique.

The latest model of the ColorRight tool (now known as the "ColorRight Pro") has a somewhat different design. I don't have one, nor do I have the instructions that come with it, so what follows is strictly from information on the ColorRight Web site.

It is apparently specifically intended for "from the camera position" measurement - in any case, that is the only modality (so far as I have seen) described on the ColorRight Web site.

The front of the ColorRight Pro is a roughly-hemispherical dome, evidently translucent (its appearance in the photos is evocative of the translucent appearance we often see in parts molded of Lexan.) A part of the hemisphere appears to be blanked by black paint or the equivalent.

What is the object of this design? We are given no hints by the manufacturer.

One might conjecture that perhaps ColorRight has recognized that the acceptance of the incident illumination on the camera position is the actual working principle (for better or worse) of at-the-camera white balance measurement, and has realized that the modest acceptance angle of the prior ColorRight tools is not ideal for that ploy. Thus, we might think that the dome is intended to broaden the acceptance pattern (perhaps substantially).

What about the blanked zone?

Well, if we were actually interested in the illumination on the camera (in its own right), then the light reflected from the ground in front of the camera is as legitimate a contributor to that as any other light. But in fact we are (presumably) only interested in the luminance there as a more-conveniently-accessed proxy for the illumination on the subject, in whose chromaticity we are actually interested.

While in fact the light from "overhead" may well be comparable in both situations (we rely on this for the technique to have any chance of success), it is less likely that there would be consistency with regard to the light reflected from the ground in front of the two sites. Thus, perhaps, it would be best excluded from the measurement.

Just conjectures, of course.

Now, what insight are we given as to the use of this new item? On the main product Web page of the ColorRight site, we find this brief set of instructions:

1. Set Manual Focus
Make sure the Auto Focus on your camera is OFF.

2. Hold ColorRight over Lens
Make sure the Auto Focus on your camera is OFF. [Yeah, we got it.]

3. Set Custom White Balance
Press menu and navigate to custom white balance and select the photo you just snapped as to calibrate the white balance.

Well, what "photo you just snapped"? There was no mention of snapping a "photo". Maybe that was what step 2 was supposed to be.

So we can't yet comment further on where the camera might be, or where it might be aimed.

If we then go the "how to" page, we find this similar brief set of instructions:

1. Set Manual Focus
Make sure the Auto Focus on your camera is OFF. This prevents the camera from attempting to focus on the ColorRight Filter when it is placed over the lens.

2. Hold ColorRight over Lens
Hold ColorRight over your lens and snap one photo. This photo will be used as the light refernce [sic] for the custom white balance setting on your camera.

3. Set Custom White Balance
Press the Menu Button and navigate to the Custom White Balance setting. Then select the photo you've taken with the ColorRight filter as the reference.

Then, as if an afterthought:

Easy as 1,2,3! Turn auto focus back on and you're ready to start shooting with perfect white blance [sic].

So now, taking a "photo" is included - but no information on where the camera is to be placed, or where it is aimed, or how the device should be oriented (the black mask is of course not symmetrical).

But of course these aren't the real instructions.

On that same page, we have a nice video tutorial by the affable Peter Gregg. It includes this interesting passage (for the second part of "Step 2"):

"Aim the ColorRight Pro and the camera together at the source of light, which is right around where the camera is."

I'll leave parsing that up to the reader.

Another interesting passage is found on the FAQs page:

• The ColorRight Pro seems to produce better results than the ColorRight Classic, why is that?

They are both capable of producing almost perfect results, everytime [sic]. However, the ColorRight Pro is easier to use in practice.

Ah, so that's it. They both produce almost perfect results, every time. But "the ColorRight Pro seems to produce better results than the ColorRight Classic". Perhaps this is a very fine distinction in "almost perfect".

Now as to the notion that "the ColorRight Pro is easier to use in practice", I not sure I recognize how that would be, but of course I have never used one.

Well, this concludes our report on recent scientific advances in white balance measurement.

Best regards,

Doug

[Andrew Rodney]

In order to compensate for a difference in the chromaticity between (a) the illumination on the scene and (b) the chromaticity of the white point of the color space in which the image will be conveyed, we need to know the chromaticity of the illumination on the scene.

Before I even go past the first paragraph, how can this device (and many like it) measure or produce information about the chromaticity of the illumination?

I was unable to imagine how a measurement frame taken in this way would consistently tell us the chromaticity of the illumination on the subject.

Me too!

[Doug Kerr]

It's not really germane to the technical issues here, but I just had to cite this passage from the current ColorRight FAQs:

• What makes ColorRight better than a grey card?

ColorRight fixes your color in camera, not in post-production saving lots of time and money. ColorRight is easier to use. ColorRight is more robust and durable. ColorRight is easier to carry and will not break apart or crumble like a grey card.

Of course, I use a gray card for in-camera color correction.

But I'm sure glad to be alerted to the prospect of my gray card crumbling! I do think the ones made of rye bread are less susceptible to that than the whole-wheat ones, which might be the kind Drew has in mind.

Carla says the gluten-free ones might be more of a problem as well. And she would have put a comma after "post-production".

Best regards,

Doug

[Andrew Rodney]

If you shoot raw, its not fixing anything (cause nothing is broken). It might produce a useful bit of metadata, but that’s a far cry from fixing something. And of course, in this raw workflow, its not affecting the raw data one bit.

[Asher Kelman]

Originally Posted by Andrew Rodney

Before I even go past the first paragraph, how can this device (and many like it) measure or produce information about the chromaticity of the illumination?



Me too!

Andy,

What neither you nor Doug seem to appreciate, are two important facts of life inherent in this product:

1. Most money in photography is made not buy selling one's own worthy pictures, but by marketing "stuff" to myriads of vulnerable camera enthusiasts.

2. This new invention "ColorRight" actually projects a virtual sheet of multidirectional light sensors over the subject, so that explains the "perfect" rendering of Chromaticity. The brilliance of this has amazed me. More so it can be leveraged using "fact of life" #1!

Asher

[Doug Kerr]

Hi, Asher,

Originally Posted by Asher Kelman

Andy,

What neither you nor Doug seem to appreciate, are two important facts of life inherent in this product:

1. Most money in photography is made not buy selling one's own worthy pictures, but by marketing "stuff" to myriads of vulnerable camera enthusiasts.

2. This new invention "ColorRight" actually projects a virtual sheet of multidirectional light sensors over the subject, so that explains the "perfect" rendering of Chromaticity. The brilliance of this has amazed me. More so it can be leveraged using "fact of life" #1!

I well understand number 1, but as for "appreciate", that's another matter altogether.

As to number 2, all I can say is "the fewer the higher, why naturally".

Best regards,

Doug

[Asher Kelman]

Originally Posted by Doug Kerr

As to number 2, all I can say is "the fewer the higher, why naturally".

That quote comes from where?

Asher

[Doug Kerr]

Hi, Asher,

Originally Posted by Asher Kelman

That quote comes from where?

My buddy from years ago, Phil Way, used to say that when presented with a nonsensical proclamation!

Phil has been for many years the proprietor (initially with his wife, now ex) of a mobile radio service shop called "Two-Way Radio". He was Cleveland's answer to Will Thompson.

Best regards,

Doug

[Asher Kelman]

Fact of life # 2 presupposes that a large enough array of multidirectional sensors placed in front of the scene to be photographed would enable us to sample the chromaticity and luminosity of the light of different incident illuminations directed towards parts of the subject. Then one could, theoretically, adjust the entire picture, detail by detail based on that collected data.

Asher

[Doug Kerr]

Hi, Asher,

Originally Posted by Asher Kelman

Fact of life # 2 presupposes that a large enough array of multidirectional sensors placed in front of the scene to be photographed would enable us to sample the chromaticity and luminosity of the light of different incident illuminations directed towards parts of the subject. Then one could, theoretically, adjust the entire picture, detail by detail based on that collected data.

Indeed. But one cannot "project" such an array from, for example, the proposed camera location. The sensors would have to actually lie on the surface of the subject. ("Excuse me, honey, but may I . . ."

To take advantage of these sensors being separately receptive to light with different angles of arrival, we would need to know, at each spot in the scene, the reflective characteristics of the surface (unless of course we could safely assume it to be Lambertian). We would also need to know the orientation of the surface at each point.

Then we would need to make some decisions as to how we would want such a surface rendered.

Ah, those pesky cosines!

Best regards,

Doug

[Craig Smith]

Very informative post. Thanks for all the valuable information.
When it comes to white balance, color correction and all the color management out there its good to have views on all the aspects to help inform yourself.

I think camera collaboration is a good know how to do. Ive come across a few good sites that walk you threw how to calibrate your camera, save it as a profile and use that profile every time you download images. If you shot RAW that is.

Here is one such site http://www.luminous-landscape.com/re...b-profil.shtml