Temperature and Tint

[RFPhotography]

You're missing the point entirely, Bill.  It seems you're so completely caught up in your own absolute literal world that you can't see the forest for the trees.  

[bjanes]

What is your definition and how am I using the term incorrectly?

You're missing the point entirely, Bill.  It seems you're so completely caught up in your own absolute literal world that you can't see the forest for the trees.  

Bob, why don't you just answer the question so this matter can be settled? You are not getting my point. I may not be the only one who is inflexible in their views.

Regards,

Bill

[RFPhotography]

The definition is moot.  'Accuracy' as you define it; as I have commented, more importantly as Tim and Andrew have commented, isn't a relevant metric.  There is a difference between numerically accurate and a colour match (or as close as can be achieved).  Scene referred 'accuracy' doesn't automatically correlate to an output referred match. 

Andrew's right.  We've been down this rabbit hole before.  And it isn't even the good pill.   

[madmanchan]

The limitations with a neutral / gray balance card are (1) it still doesn't tell you what the spectrum of the illuminant is, and (2) it doesn't compensate for the fact that the illumination can be (and often is) spatially-varying across the scene. 

So, you can click-WB on a neutral reference and guarantee that its appearance within the rendered image will be perfectly neutral (e.g., R = G = B). However, that by itself guarantees nothing about the appearance of the rest of the (non-neutral) colors in the image.  There are many factors that would need to be taken into account in order to get the colors in the image to be colorimetrically correct.  This is generally difficult because, as indicated above, the factors are generally not known at the time of capture.  Even if they are known at the time of capture (or can be estimated, or guessed, or otherwise provided), a perfect colorimetric matching is often impossible, due to mismatches in the way the camera "sees" and the way we (humans) see.  And last, but not least, it has been my experience that photographers generally don't want colorimetric matching as an end result.

[Czornyj]

And last, but not least, it has been my experience that photographers generally don't want colorimetric matching as an end result.

Because of colorimetric matching limitations, or rather because of currently actual trend & fashion in photography?

[PhilipCummins]

No I will not carry additional instruments. Everything that weighs has to be omitted. One problem such instruments would meet is that my shooting "biotope" these days is woodland, and there is so to speak never an unobstructed view to the light source.

IMHO just get a ColorRight Classic to take a white balance shot before a series of photos to capture a relatively accurate white balance and review on your camera on site if you trust the camera's LCD. If you needed a gray card you can use the ColorRight Classic to provide this as well by photographing it into the scene. Then in post production you can choose what white balance you prefer (as shot with custom WB, AWB, camera modes like Daylite, Nighttime etc).

Naturally it wouldn't be 100% accurate however it would be probably more accurate than post-processing it to what you remember. Even then, you could choose what white balance you preferred according to whatever criteria you wanted to make it look like what you recalled. If you needed more accuracy you'd probably need to grab a ColorChecker Passport or similar device to assist.

Reading other people's comments it could be hashed out to any number of degrees however for myself I've found a combo of the ColorRight Classic + review on-site of the image on the camera (usually right after taking the shot) itself to be beneficial for taking relatively accurate landscape photography scenes. Since this is for my own personal use I prefer accuracy over what looks "best" or "most pleasing" so I rarely bother to modify the images afterwards unless I'm specifically printing them off.

[ixania2]

Try raw photo processor. Its white balance AUTO is so good, it's stunning every time, and a wonderful starting point.

[Hening Bettermann]

@ Eric

I'm not sure if it is me whom you are addressing with your last post, but I'll reply as if it was.

> that photographers generally don't want colorimetric matching as an end result.
The example with the white/pink houses should make it clear, that I am aware of that the colorimetric correctness as obtained by the gray card method will probably not give me the color as perceived at shooting time - not (only) due to limitations of the method that you point to, but due to its principle, as I understand it.
That said, I believe that the way in which "your" photographers want to deviate from colorimetric accuracy has quite different reasons - they want something "pleasing".

> mismatches in the way the camera "sees" and the way we (humans) see:
I think my exspectations to this match are much more modest than what you have in mind. My problem is not that I can not get a match that seems close enough to me. My problem is that I find it good enough over a wide range, and I would like something that is more precise than my memory, which we all know is a rubber measure.

@PhilippCummins

The advantage I can see in the ColorRight diffuser over the grey card is that you operate it on the camera, or at arms length. The card requires to be placed in the distance of focus. Otherwise, the same considerations apply.

@ixania2

Thanks for the tip. However, I think I'll stay with Raw Developer as the raw converter because of its deconvolution sharpening.

Good light! - Hening.

[bjanes]

The definition is moot.  'Accuracy' as you define it; as I have commented, more importantly as Tim and Andrew have commented, isn't a relevant metric.  There is a difference between numerically accurate and a colour match (or as close as can be achieved).  Scene referred 'accuracy' doesn't automatically correlate to an output referred match. 

Andrew's right.  We've been down this rabbit hole before.  And it isn't even the good pill.   

As a non-measurbator, you likely do not have the slightest understanding of "color accuracy" but are merely spouting off what you have read in the non-peer reviewed literature on the web. Imatest is a measurbators delight and you might gain some knowledge by looking at the ColorCheck documentation, which is prefaced by this comment:

The notation in this section is adapted from the Digital Color Imaging Handbook, edited by Gaurav Sharma, published by the CRC Press, referred to below as DCIH. The DCIH online Errata was consulted.

In measuring color error, keep it in mind that accurate color is not necessarily the same as pleasing color. Many manufacturers deliberately alter colors to make them more pleasing, most often by increasing saturation. In calculating color error, you may choose not to use the exact formulas for ColorChecker L*a*b* values; you may want to substitute your own enhanced values.

So you might refer some of your comments to Norman Koren, the author of Imatest. He has devoted considerable time and effort to measure color accuracy.

With regard to the ColorChecker, you might refer to Danny Pascale's article on it and the various equations that can be used to convert between color spaces that can be used for transformations of the color values. Danny also has an excellent article on color spaces, much better than the dumbed down version given by the DigitalDog. The ICC also has a white paper on measuring color accurately. When you have gained some technical knowledge, please come back for further discussion.

Regards,

Bill

[RFPhotography]

No Bill, I understand the point you're making quite well.  I also understand the theory behind it.  I don't care about the math.  Understanding the math isn't necessary to understand the theory and use it in practical application.  You disagree with that and that's fine.  I simply disagree on the necessity of such a rigid workflow in all situations.  And I disagree that such a rigid workflow is valuable in all situations.  I've said before; and if you actually take the time to read you'd realise, that I do believe there are times when as close a scene-referred match as can be obtained is valuable.  You're also, seemingly, ignoring the point that a scene-referred match doesn't correlate to an output-referred match (i.e., that the colours reproduced in the final output are a match to the colours in the original scene).  To say nothing of the brightness range in the original scene that, depending on the range, may not be able to be reproduced by any output medium.

And while I disagree with the idea that an 'accuate' WB is the best way to achieve a proper scene-referred image and that a camera profile, similarly, is the Holy Grail I reference this from "Real World Color Management, 2nd Ed" by Fraser, Murphy & Bunting:  "Camerra profiles are a bit more slippery.  We've found that even with the best camera profiles, we'll still have to make significant edits to tone, and when camera metamerism rears its head, to color too."  Further, the authors continue to state that a good camera profile is merely a 'nudge in the right direction'.

Let me ask you this, Bill.  Given your strong preference for scene-referred colour 'accuracy', how did you ever operate in the film world?

[bjanes]

I do believe there are times when as close a scene-referred match as can be obtained is valuable.  You're also, seemingly, ignoring the point that a scene-referred match doesn't correlate to an output-referred match (i.e., that the colours reproduced in the final output are a match to the colours in the original scene).  To say nothing of the brightness range in the original scene that, depending on the range, may not be able to be reproduced by any output medium.

You keep talking about scene referred images, probably taking up the subject from the DigitalDog, who likes to use the term for obfuscation. However, I’m not certain that you know what scene referred actually means. A scene is defined in ISO 22028-1 as:
 
“scene: spectral radiances of a view of the natural world as measured from a specified vantage point in space and at a specified time.
 
A scene may correspond to an actual view of the natural world or to a computer-generated virtual scene simulating such a view.

A scene-referred image is an image where the image data is an encoding of the colors of a scene (relative to each other), as opposed to a picture of a scene. In a picture, the colors are typically altered to make them more pleasing to viewers when viewed using some target medium.”


The definition of spectral radiance from Wikipedia:

"Spectral radiance expresses radiance as a function of frequency (Hz) with SI units W•sr−1•m−2•Hz−1 or wavelength (nm) with units of W•sr−1•m−2•nm−1 (more common than W•sr−1•m-3). Radiance is the integral of the spectral radiance over all wavelengths."

Thus a scene referred image is in linear gamma and is essentially HDR for most normal outdoor scenes. Further the scene color data are determined by the actual wavelengths of light reflected from the subject. The raw file is scene referred within the limitations of the digital capture. If we want to reproduce the appearance of the scene, we have to know something about the illuminant, and this involves white balance.

As discussed at length by Karl Lang in Rendering the Print, rendering of a scene referred image is largely concerned with mapping the extended luminance of a scene to the limited dynamic range of the output medium. This may be accomplished to some extent with sigmoidal tone curve, but more sophisticated techniques are needed for HDR images. Color may also need to be remapped, but to a much lesser degree if one is using wide gamut displays and printers. With color, the saturation is often adjusted, but hue changes are not generally desired. If we had a display capable of high dynamic range and with a large color gamut, the scene referred image with proper white balance would reproduce the scene. Viewing conditions of the image are also important and are taken into account by CIECAM02. With most existing output devices, scene referred data is not pleasing, and rendering into an output space is performed. The thrust of my quest for accuracy is to obtain an accurate scene referred raw file. From there, one may make further adjustments to achieve the desired effect.

And while I disagree with the idea that an 'accuate' WB is the best way to achieve a proper scene-referred image and that a camera profile, similarly, is the Holy Grail I reference this from "Real World Color Management, 2nd Ed" by Fraser, Murphy & Bunting:  "Camerra profiles are a bit more slippery.  We've found that even with the best camera profiles, we'll still have to make significant edits to tone, and when camera metamerism rears its head, to color too."  Further, the authors continue to state that a good camera profile is merely a 'nudge in the right direction'.

I think that Fraser et al were referring to making ICC profiles for cameras. These are generally valid only under standardized environments such as in the studio or laboratory where the illuminant can be controlled. In all cases, some type of profile is needed to convert the XYZ values produced by the camera to the working space such as ProPhotoRGB. A 3x3 matrix is generally used, and further modifications with look up tables may be involved as with the DNG profiler. Some insight into how this is done in Adobe applications can be gained by studying the source code of the DNG SDK.

Regards,

Bill

[RFPhotography]

Answer the question that was put to you:  Given your strong preference for 'colorimetric accuracy', how did you ever operate in the world of film photography?

[bjanes]

Answer the question that was put to you:  Given your strong preference for 'colorimetric accuracy', how did you ever operate in the world of film photography?

Funny that you should make such a request for me merely to answer your question (which was off topic and irrelevant to the discussion, but presumably designed to trip me up), since I've been asking the same from you with no success, even for relevant questions.

Much of my film work was with black and white, so colorimetric accuracy was not relevant, but one might put a yellow filter over the lens.

Most of my color work was with chromes, and Kodachrome 25 was my long time favorite. Variations in processing were out of the question with this film and you had to get exposure and color balance correct in camera. Of course, white balance was critical. For tungsten (3200K), an 80A filter could be used. For minor corrections warming filters could be employed, but I never tried CC filters. At that time my knowledge of colorimetry was meager and of limited importance to me since I could not evaluate color except by eye and I couldn't control it other than to use proper technique. I occasionally used pre-flashing to reduce contrast.

For photo-microscopy, we often used Ektachrome EPY with a one stop push to increase contrast. When Velvia came out, that was my favorite film for photo-microscopy and landscape because of its saturated colors. Velvia was not good for portraits with Caucasian skin, and Kodachrome or Astia was better for portraits. Kodak always sought to produce accurate colors and their films fell out of favor with some who preferred the increased saturation of the newer films. As you and others have noted, pleasing color is often preferable to accurate color. It depends on what one is trying to achieve. Generally, one often wants more saturation, but hue shifts are usually undesirable. With digital, increased saturation is easily accomplished in post, but is is more difficult to correct hue shifts. That is why Imatest reports color difference with and without correction for chroma (saturation).

Now, I invite you to answer my questions on what is wrong with my evaluation of color accuracy using Imatest rather than saying it is flawed and irrelevant. I won't hold my breath while awaiting your answer.

Regards,

Bill

[madmanchan]

Because of colorimetric matching limitations, or rather because of currently actual trend & fashion in photography?

Simply preference (like film).

[RFPhotography]

Now, I invite you to answer my questions on what is wrong with my evaluation of color accuracy using Imatest rather than saying it is flawed and irrelevant. I won't hold my breath while awaiting your answer.

Regards,

Bill

You never mentioned Imatest until a few posts ago in this thread.  You've not indicated it's a part of your workflow.  You've indicated that an 'accurate' white balance is what's required for colorimetric accuracy.  That's what I and others dispute.  And you didn't answer the question I put to you.  You typed a lot but didn't answer the question.  In point of fact you confirmed what Andrew, Tim and others have alluded to all along.  Done.  Finished. 

[PhilipCummins]

The advantage I can see in the ColorRight diffuser over the grey card is that you operate it on the camera, or at arms length. The card requires to be placed in the distance of focus. Otherwise, the same considerations apply.

Yes, it's a Through-the-Lens device. The central portion diffuses the light entering the camera lens (ie, placed so it temporarily replaces where the plastic lens protector would normally be), switch auto-focus off and then set the white balance on the camera itself (I have a handy button that speeds this process up considerably without needing to switch auto-focus off). If used as a grey card it merely needs to be in the scene with enough pixels to evaluate your own white balance, not necessarily with accurate focus. There's the ColorRight Pro (and Pro Max) which apparently has better transmissive capability with a dome design to capture light from multiple angles however I find it fairly bulky (though probably not too bad for DSLR owners).

[bjanes]

You never mentioned Imatest until a few posts ago in this thread.  You've not indicated it's a part of your workflow.  You've indicated that an 'accurate' white balance is what's required for colorimetric accuracy.  That's what I and others dispute.  And you didn't answer the question I put to you.  You typed a lot but didn't answer the question.  In point of fact you confirmed what Andrew, Tim and others have alluded to all along.  Done.  Finished. 

... if I photograph an X-Rite color checker and render it into a defined color space such as ProPhotoRGB, I can compare the observed values in the file to those measured from the target by a laboratory grade spectrophotometer and determine the accuracy of the rendering and give a mathematical analysis. For more practical work, one could assume that the color checker is reasonably accurate. It helps to know the white balance for the illumination used to take the photograph. What is subjective about this?

Bob,

Not only are you very thin skinned and argumentative, but your reading comprehension and ability to draw conclusions from the text is limited. My initial post in reply to the DigitalDog's comment about accuracy is quoted above. All Imatest ColorCheck does is to compare the values of the rendered image to the nominal values of the ColorChecker and perform related calculations. White balance is necessary for this process. I later invoked Imatest to indicate that Normal Koren, the author of Imatest, thought that this methodology was sound.

Since you misinterpret what I said and won't respond to my questions and assertions and a meaningful discussion is not possible, I agree that continuing this thread is pointless. I have looked at your website and read some of your posts and conclude that you are knowledgeable, but I don't know why do you have to prove that you know more than others? You need to read this book.

Regards,

Bill