Want a RAW histogram?

In a recent thread on the Luminous Landscape forum, Guillermo Luijk (who contributes here also) described his experiments with a custom WB setting that would cause the camera's jpg histogram to more closely resemble the distribution of tones in the RAW file. (A brief explanation for those who are unaware of the problem or what causes it: The camera histogram when shooting RAW is derived from a jpg embedded in the RAW file for display purposes. Part of the process of creating that jpg includes white balancing, which is done by digitally boosting the red and blue channels because the sensor is less sensitive to these colors than to green. A sunlight WB, for instance, might call for multipliers of red x 1.9 and blue x 1.35. This boost can cause one or both of these channels to be clipped in the jpg even though they weren't clipped in the RAW. This can be a problem in older cameras that have only a luminousity histogram which most closely resembles the green channel, causing the user to be unaware of clipping in the red or blue. It is also a problem even in newer models that have three color histograms if the photographer is using the ETTR method of exposing and wants to use all the available headroom while relying on highlight recovery in the RAW converter to prevent the clipping.) The solution is to use a target for a custom white balance that will produce a CWB whose multipliers are 1.0, that is, no actual boosting is done. This is the link to Guillermo's post where he describes his method:
http://luminous-landscape.com/forum/in​dex.php?showtopic=2225​0&st=0
The idea is not entirely new, Nikon users did something similar a couple years ago. However, what is of interest to us is that most of the participants in the thread are Canon people and some of them created CR2 files for making the CWB and made them available for downloading. The thread itself is quite long and wanders into several other topics, so to save you time I will list here those links:
For the XT/350D - http://www.guillermolu​ijk.com/download/uniwb​350d.cr2
For the 40D - http://www.cryptobola.​com/PhotoBola/WB_00040​.CR2
For the 1DMk2 and the 1Ds - http://www.visual-vacations.com/images/2​008/

You should be aware that there are certain disadvantages to using this CWB. First, reviews on your LCD will have a green cast, the "native" color of the unbalanced image. Anybody looking over your shoulder while you chimp will tell you that your camera is broken. Second, in experiments with my 40D I have found that if the green channel is overexposed and clipped but the red and blue channels are not (because they were not boosted), a single clipped channel is not enough to activate the blink warning, so watch the green histogram. And, of course, you will have to spend time correcting the WB in your converter.Other than this, this appears to be a very useful tool for the RAW shooter.

Is the link to the 40D correct? It just looks like an image of a blurry Photoshop screenshot?

Yes, that's it. It's a photo of the proper color displayed on a monitor. It is intentionally blurred in order to smooth out the screen's pixels. Since the camera uses only the central 9% circle for setting a CWB, what is captured is enough.

So, how do you use it? Do you print it? Also, is it supposed to work regardless of actual light color temperature?

With a card reader put it on a CF card. The naming convention is in accord with that used by the 40D and the camera will recognize it. Use it to make a custom white balance the same way you would with an image you shot for that purpose,

I forgot to mention in my first post that when you open a RAW shot with this CWB in your converter it will of course be green just like the LCD display and you will need to correct the WB by the usual methods - eyedropper, preset or Auto.

Ah, that makes sense. I have gotten my controlled WBs with a card, so I was imagining printing this out. That is why the extra material threw me.

This is interesting. My question vis-a-vis actual color temperature was, was it found that the camera applies the same boosts to the red and blue channels across the range?

It is also interesting, since it appears that the 40D boosts the red and blue far less than the XT. In fact in Photoshop, the green and blue are identical and the red channel is lower.

I suppose I should stop trying to be lazy and actually read through the thread.

The Hardcard wrote in post #4773129
Ah, that makes sense. I have gotten my controlled WBs with a card, so I was imagining printing this out. That is why the extra material threw me.

This is interesting. My question vis-a-vis actual color temperature was, was it found that the camera applies the same boosts to the red and blue channels across the range?

I suppose I should stop trying to be lazy and actually read through the thread.

No, the multipliers vary according to the temperature of the ambient light. Sunlight is blue, so to counteract that the red is boosted more. Tungsten light, OTOH, is red, so the blue gets a bigger push

Oops. Yes, I am still approaching it not quite right. I now see better how this is working. It limits the camera's distortion of the channels necessary for the the image to appear the correct color to the viewer.

I am still a little confused by how nearly perfectly gray the 40D sample is.

Nothing for a 5D?

ErikM wrote in post #4773296
Nothing for a 5D?

Yes, there is already one RAW file for the 5D, and 30D is coming tomorrow from a 30D user. Feel free to download them from the original article page (it's in Spanish, just head for the 'DESCARGA DE ARCHIVOS RAW UNIWB' area): http://www.guillermolu​ijk.com/article/uniwb/​index.htm.

Supported cameras:
- UniWB350D.cr2
- UniWB20D.cr2
- UniWB30D.cr2
- UniWB40D.cr2
- UniWB5D.cr2
- UniWB1Ds.cr2
- UniWB1DMkII.cr2
- UniWBD300.nef


The concept of this UniWB is flawless: if we achieve 1.0 multipliers all the work we could do is done. Now it's a matter on how actually the JPEG development process is performed on each camera model that the final histogram is the same as the RAW histogram.

But a lot of improvement exists, I am using it on my 350D and when I switch to 'Daylight' or 'Tungsten' I feel a bit lost wondering how much of the information blinking in my display is actually blown. Working in UniWB, although not perfect, is far more accurate.

BR