I dunno, the analysis is really fuzzy. He takes the resolution of our vision at the center then spreads it over our full 180 degree perspective. But the resolution at the center is incredibly narrow. We can inspect a scene at a high density, but that's an interactive process. As long as we allow for an interactive process, heck, we could walk around, swivel our head around, pick up objects...

Really our eyes are just nothing like cameras. They are more equivalent to a video camera, at which point the comparisons can be meaningful. And the resolution of video doesn't need to be very high in comparison to still images. Aren't digital movies distributed at just a couple megapixels?

I like it.

But bear in mind that our eyes are actually quite low-res. Its our brains that bat our eyes around to build up that complete picture...

I do *wish* there was some sort of head-mounted camera that could capture the complete immersion that us humans have, in terms of our visual ability.. Some time off, I'm guessing... Years, decades - it doesnt really matter.

I want it *now*.

---* Bill "Blogging in the Nude" Buchan...

Yes, the eye works very differently, especially in how it pieces together the image. But in terms of trying to understand the camera specs we'd need in order for a camera to capture what the eye does, the comparisons are valid.

Also, I was thinking about that 800 ISO number. It seems quite low. The eye at night can see much more than an ISO 1600 camera, so why isn't the ISO number higher? I get the feeling we're measuring some technical ability which doesn't correlate to sensitivity of capture...

hmm... The analysis does seem to be a little off. He almost dizzied me with these weird units of measurement, though -- candela? steradian? I don't think so...

The aperture value of f3.2 also seems far too slow -- http://en.wikipedia.org/wiki/F-number claims that it's f2.1 at the maximum. Even that surprises me. The eye's "shutter speed" must be very low, and I don't understand how that would work.

I am left with more questions than answers.

Regarding the supposedly low iso.. don't forget that the eye will 'accumulate' multiple frames (in essence it combines the last second or so of picture so that we can see edges etc). We also have an inbuild 'shadows highlights' feature than allows us to extract meaningful information from a pretty quantised 'photon by photon' image.

Tim: you are right, but there's a discrepancy. When computing the megapixels, Clark came up with an effective resolution that applies after all of the scanning-and-composing that the eye does for us. But then when computing ISO, he used the raw sensitivity, rather than the effective sensitivity that results from the accumulation over time the eye and brain do for us.

Waw, that is some lot of pixels, but I don't believe it matters how many pixels you have to record data but what matter is how much data can be recorded about one pixel.
It's really the range of light that can be recorder about one pixel that matters when comparing why a picture you take is different from what you see. If a camera can record details on a range of say 5 F stops, a human eye may record, some people say 10 F stop an other would say even 20 F stops. So what to your eye is visible detail, to the camera sensor is just heavy under or over exposure. To get somewhat closer to the human vision there is a technique a photographer can employ in which multiple bracketed exposures are being taken of the same spot and then later combined in with some software, and if this should theoretically work, practically it dose not because there are other problems too.
Another problem is that even if digital sensor or film is capable of recording say 5 F stops, a typical monitor or printer or paper surface is not and will show you even less detail.
Another problem is how you store the picture, if you take raw then you got more details recorded, if you take JPEG then on 8 bit per channel like the typical RGB then the amount of detail is yet again less then what the best sensor, say a full frame 16 Mpx, out there will record.
Is there something to do for the picture to look more like the original scene you saw? One thing would be to take the picture when there is not too much contrast like mid day sunny day, but rather overcast even illuminated day or use artificial light to illuminate the parts that are underexposed or use reflectors to direct the natural light on parts that would typically record underexposed, or use software like Photoshop to correct exposure, contrast and color problems etc.