Timo and the 12 bits

Timo Autiokari published a web page, 12 bit Acquire Devices and Gamma, that claims to describe 12-bit image acquisition. I believe that Timo does not understand the fundamentals of nonlinear coding; see Concerning Timo Autiokari and Timo incident - The technical argument. This note clarifies Timo's 12 bit page. Prior to your reading this page, I recommend that you read the Gamma FAQ. Perhaps you'd also like to read about Linear and nonlinear coding; that document forms sort of a gentle introduction to the Gamma FAQ.

Gamma is ordinarily expressed as a positive number with a value of approximately 2.5. A CRT monitor responds to voltage V by producing an intensity proportional to V raised to the power 2.5. Gamma correction is the process of applying a power function of about 1/2.5 - that is, about 0.4 - to intensity, to produce a code value suitable to drive a CRT monitor. As detailed in the Gamma FAQ, the amazing thing about a CRT is that its response is almost exactly the inverse of the lightness sensitivity of human vision. So, by gamma correcting, we are making the code values approximately perceptually uniform. This is a good thing.

Timo says at the opening of the page,

Imaging devices that have greater bit-depth than the 8 bit are able to bury the monitor gamma compensation into the byte values of the image before the image is converted into the common 8 bit/color format.

Timo says, "gamma compensation." To conform to industry-standard terminology, he should say, "gamma correction."

Timo continues,

The added accuracy will be infiltrated into the image gamma.

In the English language, the word "infiltrate" is pejorative. Here, the incorporation of gamma correction is a good thing, so the word "infiltrate" is misleading.

Timo says in the second paragraph,

Because the inverse gamma calculation is done at higher accuracy integers, there will be less round off errors and much smoother quantization.

Timo says "inverse gamma." More properly, he should say "gamma correction." Because gamma is a positive value approximately 2.5, application of gamma is application of that power. It is normally done by the CRT. Gamma correction is application of a power about 0.4.

Timo continues in the second paragraph,

Doing so results much better quality images (but for the uncalibrated CRT monitor only) than is possible to produce from linear image files (because the images will be in the normal 8 bit/color space at the time of publishing).

Timo says "uncalibrated CRT monitor." By "calibrated," Timo means "linearized." But the CRT monitor itself has no means to compensate its nonlinearity. This can be done by loading a suitable lookup table into the graphics card that drives the monitor. You can load a lookup table into which is embedded a 0.4-power function.

Main thread:

Timo thread: