🚀 go-pugleaf

Rags,

In a raw capture, there are no color channels�the raw file simply records the luminance value from each sensor site, along with the color filter array map that indicates which color filter is applied to each sensor site. The luminance information is recorded as digital counts, typically 12-bit (0-4096), that are directly proportional to sensor voltage and hence directly proportional to the number of photons hitting the sensor. The digital values are linear with respect to the number of photons that impinge on the sensor�that's what linear gamma is.

The demosaicing uses the color filter array map to produce color from the grayscale. The gamma correction turns the linear tone curve into one that more closely matches our eyes' response to the same photon count.

If you make a gray ramp from 0-255, and assign a gamma 1.0 profile to it, you have a close representation of how the sensor sees incoming light, and you'll notice that most of the RGB counts are devoted to highlights.

Assign a gamma 2.2 profile to the same gray ramp and you have a close representation of how the human eye responds to the same photon counts�our eyes are much more sensitive to small differences in brightness in shadows than they are to small differences in brightness in highlights.