"Correct" white balance

[Les Olson]

Have I got it right; Nikon doesn't share their "codes" regarding color etc. with other software but their own (Capture NX- D)? So if one want Nikons "own" colors, one have to use their software... If so, then I take it for granted that software like PS CC / ACR / Lightroom make their own Nikon profiles by testing/"guessing". By using color checker like for instance X- Rites: one is making profiles by X- rite definition of color, and not Nikons?

It is not that Nikon's definition of "red" is different from X-rite's.  Colour is defined, for the purposes of printing and so on, by the tri-stimulus model - hue, brightness and chroma or saturation.  Each of those three values can be specified as a number and the resulting colour is precisely and unequivocally described.  Hue, in turn, is the RGB values.  Hue, chroma etc do not exist in the real world: the photons have wavelengths and there are more or less of them and that is it.  The sensor just captures photons, and values for hue, brightness and chroma have to be created by the image processing engine to create the recipe for the colours on a screen or on paper.  What the image processing engine does varies, and that is what creates Nikon colours vs Adobe colours. 

The value of a profile based on a colour card is that the card has specified RGB values for each square and standardised brightness and chroma.  A colorimeter can measure the RGB values for each patch on your screen, then software works out how to adjust the display so the output has the same RGB values as the colour card.  Then your printer can use the same RGB values and get the same colour as the card. 

The catch is that the tri-stimulus model assumes that hue, brightness and chroma do not interact, and in the human visual system they do, so accurately reproducing the colours on the card with standardised brightness and chroma does not guarantee that colours with different brightness and chroma will be "accurately" - ie, as we see it - reproduced.

This is also why ETTR is sometimes a bad idea.  When you reduce brightness (misleadingly called "exposure" in Lightroom) in software the program does not change all the values equally: mid-tones are changed most and the extremes least.  That causes shifts in perceived colour.  There is an example at http://nikongear.net/revival/index.php/topic,5905.msg95734.html#msg95734

[Ethan]

The technical term to search is "color constancy". There is a very elegant example at http://persci.mit.edu/gallery/checkershadow

Let's be here serious. What Les Olson said here and in the Pixel Shift thread is correct. However, I would not call it "color constancy" but Color Balance or Color Harmony in the sense of achieving the totality of an image stimulus.

Without going into a full fledged tech discussion on colour, the white balance part of it used by in camera software or digital software plays with the Brightness side of color v Hue and Saturation. In other words, it does not matter as the final judge is your own eyes. Perception is key and in camera alone cannot achieve it and white balance is the poor man solution.

When it comes to White Balance, I would not give a monkey at shooting stage unless I am shooting technical where things get tricky and we have been doing our colour correction using calibrated Sony Gamma monitors and VectorScope. Today if you are serious about colour, Photoshop levels and cureves and whatnot simply do not cut it. I appreciate entirely that the cost of a gamma monitor and vectorscope is beyond the needs and the means of run of the mill photography. We use it strictly for commercial shoots where colour harmony is critical. I also use it for standard shoots in case I am sitting in front of the station.

To summarize, White balance is as irrelevant as whatnot and camera manuf did not have any choice but come up with this one and Photoshop came up with curves which means reliying on your eyes with no reference hence the need to use a gamma monitor or you are toast.

I reiterate, Gamma stuff is for very high end and the world continues to revolve without it. As for white balance, it is a stop gap as whatever you do. it is only your eyes who are the judge as everything you do is perceptual!

How many of you changed the gamma in their Photoshop application and their colour settings and without being pedantic, do you know why and how you should change the settings. 99% of users simply accept Photoshop crap defaults.

Which brings me to another subject related to Photoshop. Do you have the correct Photoshop settings. Nowhere and nobody will ever reveal their settings.

[TedBaker]

The catch is that the tri-stimulus model assumes that hue, brightness and chroma do not interact, and in the human visual system they do, so accurately reproducing the colours on the card with standardised brightness and chroma does not guarantee that colours with different brightness and chroma will be "accurately" - ie, as we see it - reproduced.

Isn't this what the different characteristic curves of each layer of traditional colour film attempts to address? Successfully or not is another question. That's to say that for any given "brightness" the ratio of red, green and blue "recorded" will be different.

[Frode]

It is not that Nikon's definition of "red" is different from X-rite's.  Colour is defined, for the purposes of printing and so on, by the tri-stimulus model - hue, brightness and chroma or saturation.  Each of those three values can be specified as a number and the resulting colour is precisely and unequivocally described.  Hue, in turn, is the RGB values.  Hue, chroma etc do not exist in the real world: the photons have wavelengths and there are more or less of them and that is it.  The sensor just captures photons, and values for hue, brightness and chroma have to be created by the image processing engine to create the recipe for the colours on a screen or on paper.  What the image processing engine does varies, and that is what creates Nikon colours vs Adobe colours. 

The value of a profile based on a colour card is that the card has specified RGB values for each square and standardised brightness and chroma.  A colorimeter can measure the RGB values for each patch on your screen, then software works out how to adjust the display so the output has the same RGB values as the colour card.  Then your printer can use the same RGB values and get the same colour as the card. 

The catch is that the tri-stimulus model assumes that hue, brightness and chroma do not interact, and in the human visual system they do, so accurately reproducing the colours on the card with standardised brightness and chroma does not guarantee that colours with different brightness and chroma will be "accurately" - ie, as we see it - reproduced.

This is also why ETTR is sometimes a bad idea.  When you reduce brightness (misleadingly called "exposure" in Lightroom) in software the program does not change all the values equally: mid-tones are changed most and the extremes least.  That causes shifts in perceived colour.  There is an example at http://nikongear.net/revival/index.php/topic,5905.msg95734.html#msg95734

Thank you for an interesting reply, Les! Clearly, there isn't a "quick fix" here..... 

I've never thought of the effect of reducing the Exposure in the way you describe.

[Frode]

Let's be here serious. What Les Olson said here and in the Pixel Shift thread is correct. However, I would not call it "color constancy" but Color Balance or Color Harmony in the sense of achieving the totality of an image stimulus.

Without going into a full fledged tech discussion on colour, the white balance part of it used by in camera software or digital software plays with the Brightness side of color v Hue and Saturation. In other words, it does not matter as the final judge is your own eyes. Perception is key and in camera alone cannot achieve it and white balance is the poor man solution.

When it comes to White Balance, I would not give a monkey at shooting stage unless I am shooting technical where things get tricky and we have been doing our colour correction using calibrated Sony Gamma monitors and VectorScope. Today if you are serious about colour, Photoshop levels and cureves and whatnot simply do not cut it. I appreciate entirely that the cost of a gamma monitor and vectorscope is beyond the needs and the means of run of the mill photography. We use it strictly for commercial shoots where colour harmony is critical. I also use it for standard shoots in case I am sitting in front of the station.

To summarize, White balance is as irrelevant as whatnot and camera manuf did not have any choice but come up with this one and Photoshop came up with curves which means reliying on your eyes with no reference hence the need to use a gamma monitor or you are toast.

I reiterate, Gamma stuff is for very high end and the world continues to revolve without it. As for white balance, it is a stop gap as whatever you do. it is only your eyes who are the judge as everything you do is perceptual!

How many of you changed the gamma in their Photoshop application and their colour settings and without being pedantic, do you know why and how you should change the settings. 99% of users simply accept Photoshop crap defaults.

Which brings me to another subject related to Photoshop. Do you have the correct Photoshop settings. Nowhere and nobody will ever reveal their settings.

Thank you Ethan!

Interesting!

I've adjusted the standard settings in Photoshop and I also run a calibration on my screen every now and then. Of course, far from optimal from a professional point of view. Though, I will say that I get fair enough prints compared to what I see on my screen. Though, always searching to optimize  .

I appreciate your detailed response, Ethan. Always something to learn!

[Les Olson]

Isn't this what the different characteristic curves of each layer of traditional colour film attempts to address? Successfully or not is another question. That's to say that for any given "brightness" the ratio of red, green and blue "recorded" will be different.

There is a phenomenon that if you increase luminance the perceived hue changes (the Bezold-Brucke effect), but it is complex - the shift is towards blue if the stimulus is below 500nm (cyan) but towards yellow if it is above 500nm, so it is hard to see that colour film could adapt to it.  There are other perceptual effects film could not modify that invalidate the assumption that the tri-stimuli are independent - in particular, if you increase luminance perceived saturation increases (Hunt effect; this is why colours in photographs taken on bright sunny days are not as vivid as we perceived them), and if you increase saturation the perceived luminance increases (Helmholtz-Kohlrausch effect; this is why women wear bright red lipstick).   

[TedBaker]

the shift is towards blue if the stimulus is below 500nm (cyan) but towards yellow if it is above 500nm, so it is hard to see that colour film could adapt to it. 

I was not suggesting that it would adapt to the colours in the scene (the stimulus), but the characteristic curves does show a different bias in the shadows, mid tones and highlights. It would make sense to pick the "best bias" that results in the most pleasing result as I agree I don't see how it can adapted to the colours in the scene. It must be there for a reason, it was my understanding it's there to give a more perceptually correct, or pleasing result.

[Les Olson]

Frode, as someone living in a winter country, you should be familiar with snow appearing blue instead of white in shadows ... The camera simply tells the truth and we believe our erratic eyes.

The blue snow in shadows is not an illusion: the snow is really blue because it is lit by the blue skylight.  That is why the camera also captures it as blue.  The principle was demonstrated by Goethe as part of his (futile) attempt to show that Newton was wrong about colour; Newton said that shadows were the absence of light, so they must be black. 

[Bjørn Rørslett]

As I said, the camera tells the truth. In the old days, colour film did the same re snow in shadows (however film lied about a lot else). As long as we had white snow as the reference within the field of view, we "saw" more white than blue, however. Thus the perceived whiteness of the snow all over was a sensory illusion.

[David H. Hartman]

This has to do with the fact that shadow areas tend to have the complementary color of the sunny parts.

The sun is warm 4800-5600K or something like that. The sky is blue and that's the light that fills the shadows. Maybe 11000K... unless ...it's smoggy then the shadows are filled with warm light reflected off the smog in the later afternoon. You point the color meter west in the afternoon and the color temp is 5000K or so. You point the color meter east and it's WTF? it's still 5000K. I'm just guessing here but I did this once and thought open up your mind Dave. Look at that light source. Our brains fool us but as photographers they can't do it totally and they can't do it all the time.

I see the color of various light sources but not fully. That's why I like to under correct WB because when I'm aware my brain can only partly fool me. I mean I see yellow late afternoon winter light as yellow but not as yellow as it really is. I see the yellow light messes with the color of objects bathed in it.

In days of film one had to be careful of under exposure of shadow areas when using deep orange and red filters. The film was more sensitive to blue. I shot an old tree trunk, burned inside and dead. I wanted to show a ghost of detail of the burned charcoal inside. I shot one sheet of 4x5 normally then one and two stops over and I should have shot +1, +2 and +3 EV. Plus two barely gave me what I wanted. I was using a simplified zone system and measuring the dark inside with a 1* spot meter. The negative was hard to print. The charcoal inside was lit by only sky so the blue light had to make it's way through a Wratten #29 filter and not much did. Then it had to reflect off the shiny part of the charcoal.

When I was really green I tried a film speed test of Tri-X under tungsten light. I only got 125 ASA. The film was quite sensitive to blue light and the tungsten offered yellow, orange and red. I conservatively rated Tri-X (non-professional) at EI 200 in daylight but it was more like EI 250. Another open up your mind Dave and think this out. I rated film at the edge (not deep corner) of a 105/2.5 negative at f/5.6.

Dave Hartman

Again I'm guessing color temperatures for sunlight and skylight. 5200 or 5300K is often the standard for daylight color side film as I recall.

[TedBaker]

As I said, the camera tells the truth.

Isn't that it in a nutshell? Yes the camera tells the truth but the story needs to "shortened" or edited to match our perceptions given the limitations of the display medium, because the display medium in combination with its viewing environment is not cable of the telling the truth...

[David H. Hartman]

Women wear bright red lipstick to look aroused. Well maybe they don't think about but it makes them more interesting to who ever they want to look interesting to.

When I was in my teens young women wore pale lavender lipstick to look...  dead?  Old women wore primary red lipstick as that's what was popular during WWII.

It's all very confusing.

Dave Hartman

[David H. Hartman]

As I said, the camera tells the truth.

The truth,  the whole truth and a very slanted truth  ...but still the truth.         

Who was it that said, "Now we know that we shall never know."?

Probably a nuclear physics but in my messed up head the statement seemed appropriate.

Dave Hartman who collects photons as a hobby. 

[Forgot the smilies.]

[Les Olson]

I was not suggesting that it would adapt to the colours in the scene (the stimulus), but the characteristic curves does show a different bias in the shadows, mid tones and highlights. It would make sense to pick the "best bias" that results in the most pleasing result as I agree I don't see how it can adapted to the colours in the scene. It must be there for a reason, it was my understanding it's there to give a more perceptually correct, or pleasing result.

Yes, I think you are correct; films differed because of the manufacturer's beliefs about what people with particular uses wanted.  So Velvia was designed to give highly saturated landscape colours because that is what landscape photographers wanted. Colour print film was designed to give pleasing skin tones - unfortunately, only white skin was taken into consideration.  It was not until the 1990s that Kodak changed its film to render brown well - and even then, it is said, it was because the chocolate manufacturers complained.

[MFloyd]

"Wrong colour" can also be because of coloured reflections, such as this plane flying over a grass airfield. Colour correction is needed to neutralise the green on the subframe.

(1) without correction
(2) with correction