Sunday, July 22, 2012

Seeing Is Over-Rated, Part II

I wrote earlier on the limitations of the human eye, and why it makes for a very poor photodetector, when compared to other photodetectors that we currently have. The earlier blog entry dealt with the response bandwidth of the human eyes, and the quantum efficiency.

This time (no pun intended), we will deal with the response time, which will produce the time resolution, of the human eye. We all know that when we go see a movie, it is nothing more than a series of still-image frames, moving past us fast enough that we do not see its motion, but rather see the image as being continuous. Standard movie frames (at least till all the new advancements in movie projection) used to go at 24 frames per second (FPS). This translates to 0.04 second per frame. We also know that the human visual system holds an image for about 0.02 second. It means that anything that comes into our visual system faster than 0.02 second will not be perceived as being distinct. So the 0.02-0.04 second is roughly the time resolution of the human eye.

Now, compare this to other devices. I've listed before some typical photocathodes used in accelerators. Note the time responses for the various types of photocathodes. The worst of these are in nanoseconds. This is still order of magnitudes shorter than the human eye! One example is GaAs, which is a common photocathode use in both accelerators and photodetectors. On Pg. 25, one can see measurement of the time response. The full-width-at-half-maximum of this photocathode is of the order of picoseconds!

So the human eye is not only a bad detector in terms of its bandwidth range and also in terms of sensitivity, it is also a very SLOW detector and can't separate a series of event occurring faster than 0.02 second!



LGW said...

As much as I would like single-photon sensitive, femtosecond rise-time and a broader sensitivity bandwidth, I think the eye must at least be more energy efficient. The eye is also very pixel-dense!

The human body, on about 2000 kcal per day, requires about 100 W. I couldn't find anything on the specific power requirements of an eye, but some single-photon detectors (Si APDs from Thorlabs) require about 9 W. I would be very surprised if the eyes account for 10% of the body's power consumption!

To be completely fair, a lot of CCDs can run through USB and so probably have about 0.5-1 W of power requirement. They usually would not be faster than an eye if they had the same "pixel density" (see ) but would have much better QE. I still don't think my eyes require 0.5 W each.

Although the number of "pixels" of the eye is unrivalled, it does not have much of an angular resolution...

None said...

Before we disparage the poor human eye too much, here is one area where "seeing" with the eye is not over-rated: dynamic range. Due to the logarithmic response, the human eye is capable of processing scenes over a range of 13-14 magnitudes of brightness. This is far superior to any CCD device in existence today. This is why high-dynamic range photography using even the best digital SLR's requires "stitching" together multiple exposures - one for the shadows, one for midtones, and yet another for highlights. The eye+brain does this instantly and automatically.