[6-22] "Modulating color"

The more photography I do, the more a basic question fascinates me: "How does an image brighten?"

Take an image at -1.66 EV (1.6 stops under "proper" exposure) and one at proper exposure; what mathematical property governs increases in brightness per pixel, such that the underexposed matches the correctly exposed image?

In a failed state, the corrected image skews color (i.e. loses fidelity) as the exposure is increased.

This is unwanted, given we photographers sometimes underexpose on accident.

This is a non-trivial question because it doesn't appear straightforward how to brighten a pixel. It may depend on many factors, like light intensity, which depends on information contained in the underlying data structure (e.g. RAW).

If we move from dark->bright, we progress from rgb(0,0,0) to rgb(255...). In greyscale, only one number changes (red/green/blue are identical).

This makes greyscale less complex; but the world has color.

Say, skin has has a hue of 21 (~skin tone); we must move from dark orange, to intermediate shades of orange, eventually to white.

In essence, the red channel approaches 255 faster than green; blue moves slowest because its opposite is yellow, which we want.

When an image is computationally brightened, the computer "guesses" how a color changes (rates of rgb). This is all beyond me; to acknowledge it is crucial, because it turns out – I think – color and exposure are basically the same thing.