Checking understanding - Colour Space and WB - geek alert

[davidedric]

Hi,

Sorry yet another thread. I've spent a while getting my head around things, and I think I have it right, but I'd like to check my understanding of a couple of things.

Colour Space representation: If I look at a three dimensional colour space representation (such as the ones near the top of the Color Management : Color Space tutorial) there is a white point at the top and a black point at the bottom. Does it follow that a line (not necessarily straight!) between these points connects all the neutral gray tones, at different levels of luminance? And, as a corollary that any two dimensional slice across such a colour space will have a neutral grey somewhere "near the middle"?

White Balance: In PP software (I happen to use LR) there are two sliders in the WB controls: temperature and tint. In a two dimensional slice taken from a colour space, the (black body) temperature can be represented by a curved line (i.e. a surface in the three dimensional space) running from the red tones to the blue ones (I think this is properly called the Planckian locus). When we adjust the temperature controller we move the representation of the image along this line or surface. The edges of the colour space approximately orthogonal to this line are magenta (mid-way between red and blue) and green (the third primary). Hence, when we use the tint slider we are moving the line/surface toward magenta or green - only roughly because I think that the direction of movement changes as we move along the locus.

Have I got this right? Sorry, I know it's geeky, but I like to understand what is going on.

Dave

[revi]

Hi,

Sorry yet another thread. I've spent a while getting my head around things, and I think I have it right, but I'd like to check my understanding of a couple of things.

Colour Space representation: If I look at a three dimensional colour space representation (such as the ones near the top of the Color Management : Color Space tutorial) there is a white point at the top and a black point at the bottom. Does it follow that a line (not necessarily straight!) between these points connects all the neutral gray tones, at different levels of luminance? And, as a corollary that any two dimensional slice across such a colour space will have a neutral grey somewhere "near the middle"?

The colour space is a continuous function, so yes, there should be a curve consisting only of neutral gray points of varying intensity. Another way of seeing that: there are different ways of choosing your coordinate system, as long as you have three independent variables; RBG is one way , but in some representations, there's a luminosity axis and two colour axes. So, the points on the luminosity axis are neutral grey. Lab is an example of such a system. HSV is another: when saturation is zero, value represents the full range of neutral grays.

White Balance: In PP software (I happen to use LR) there are two sliders in the WB controls: temperature and tint. In a two dimensional slice taken from a colour space, the (black body) temperature can be represented by a curved line (i.e. a surface in the three dimensional space) running from the red tones to the blue ones (I think this is properly called the Planckian locus). When we adjust the temperature controller we move the representation of the image along this line or surface. The edges of the colour space approximately orthogonal to this line are magenta (mid-way between red and blue) and green (the third primary). Hence, when we use the tint slider we are moving the line/surface toward magenta or green - only roughly because I think that the direction of movement changes as we move along the locus.
(...)

What is important for white balance is the relative intensity of the three colour channels. As we are interested in relative intensities, we can force one to be fixed, and only need to adjust two.
What makes it a bit more complicated is that optimal correction is not a matter of just adjusting red or blue (leaving green fixed) as that would also change intensity (and we want to avoid that
as much as possible while adjusting WB). So the actual corrections are linear combinations of the three channels (mathematically speaking there's a change of base vectors involved, such that intensity becomes one of the base vectors)