Added by Simen1
September 16 , 2016
Re: Maximum Dynamic Range
In the basic principle you are right since raw files have linear values for intensities, but there are an explanation to the higher numbers. Those numbers are resolution normalized, and this shifts the DR results somewhat upwards. The short explanation for the shift is this: When reducing the resolution you are averaging pixel values. The full well levels will be the same, but averaging noise will reduce noise. The noise floor gets lower. For a 16 Mp camera the shift is 0,5 Ev. For an 32 Mp camera the shift is 1,0 Ev and so on for every doubling. All files are normalized to 8 Mp.
Another factor to consider is that this is assuming a monochrome sensor. A RGB color sensor have different sensitivities, or color filter strenghts, for the R G and B channels when shooting grey. If you look at the color response details for 5D mark IV, the green channel have double the sensitivity compared to red.
This means that neutral grey will give different raw file intensity values for the color channels up to about 1 Ev difference between red and green, and get saturated at different grey levels. In other words a larger DR. This is tested values.
If we take a step into a hypothetical discussion, we could think of a sensor with a chess board of ND filters in front of RGGB quad pixel clusters. That would increase the DR at the cost of other metrics, even without increasing bit depth in the files. This is practically a 1 shot HDR technique. This or ordinary multi shot HDR are practically unlimited in terms of DR, even if the files are limited at lets say 14 bits. A bit less hypothetical is to have RGGB with two different greens, one darker then the other. R, G1, G2 and B, to boost the DR a bit at the cost of noise.
Back to the real world. As far as I know there are one sensor with 16 bit ADC. It is the Sony made 100 Mp full frame 645 medium format sensor (53,7mm x 40,4 mm) that sits inside a couple of high end cameras. Phase One XF 100 Mp and Hasselblad H6D-100C and probably also a Pentax 645 camera sometime in the future, since its known that they made a deal with Sony to potentially use this sensor.
In my opinion the full frame market is ready for an increase to 16 bit ADC and raw files because modern large pixels have a very large full well (50k-100k electrons) combined with extremely low read noise in the order of 1-2 electrons RMS. This causes 14 bit ADC to limit the capabilities of the pixel array. Technically they could go to 15 bit, but that would make a need for another transition just a few years ahead.