sonyalpharumors

The eye doesn’t have a bigger dynamic range than a camera if you include both as a system, and not just a sensor. The eye has a total dynamic range approaching a million to one, or a bit under 20 stops. However, that dynamic range cannot be perceived at the same time – that’s something a bit under 10,000:1 or about 13 stops. To achieve the full dynamic range, the eye has to adapt by altering the iris and, for low light, some chemical changes in the retina. As it is, that 13 stops is a good match for the DR of good DSLR sensors. Some CCD sensors can manage another couple of stops above that.

However, if we include the ability to stop down a lens, which typically gives another 7 stops, the camera’s full DR is also about 20 stops. Beyond that there are ND filters (but then we can wear sunglasses).

It’s also worth noting the eye sacrifices colour perception at low light levels and is also only able to gain full resolution over a small portion of the scene – we gain an illusion of having high acuity across a scene because we “scan” it rapidly.

To be fair to the eye, we do manage this 13 stops of DR on a moving image, although I’m not sure it is strictly possible to treat these the same. As the “persistence of vision” issue shows, our visual processing has its limits. It’s known that the eye has a “fast route” into the brain that perceives potential danger much faster than that which goes through or conscious mind.

If Sony do go down the route of adding “white light” photosites to augment the colour ones, this will neatly mirror the eye’s structure where we have more sensitive rods for B&W perception with cones to add colour.

I’d also be interested to see if any manufacturer could produce a sensor which had a pseudo-random placement of photosites of different colour sensitivities. That would have the great advantage of eliminating many artefacts caused by regular patterns, such as moiree, but at the cost of being fiendishly difficult to process.

It’s also interesting to wonder what a B&W sensor might achieve. Bayer pattern filters throw away about two-thirds of the light. A dedicated B&W sensor could expect to gain better than a stop on a colour camera and have better DR (as the noise-floor due to photon shot noise will be reduced).