The Dynamic Range of the Human Eye

[Jody Neckles]

My hope is one day images aquired by digital cameras will be just like the human eye.

In researching the capabilities of the human eye I read this artical on Pro Video Coalition.

http://provideocoalition.com/index.php/jdvandenberghe/story/the_eye/

It suggests the human eye has the Dynamic Range of 20 Stops but this is *not* available simulatenously. The eye actually adjusts its sensitivity to different lighting conditions (chemical changes in the eye and adjusting the iris) and can only see 6.5 Stops simulataneously. This range is known as the Brightness Adoption Level.

I find 6.5 Stops difficult to believe. Am I understanding this correctly?

So what would be the Brightness Adoption Level be of the Epic which is supposed to be 18 Stops with HDRx™™?

 

[jbeale]

If you are talking about "not simultaneously" available dynamic range, there's no limit on the upper range of any camera since you can always add external ND filters to take pictures of the surface of the sun, etc However, the dynamic range of HDRx is actually all available in the same frame (maybe not exactly simultaneously, see "magic motion", but in the same frame).

 

[Kalani Prince]

Right now, I'm seeing the world at about f.045 : ) maybe your brain trashes extra data as you change focal points.

 

[Jody Neckles]

If you are talking about "not simultaneously" available dynamic range, there's no limit on the upper range of any camera since you can always add external ND filters to take pictures of the surface of the sun, etc However, the dynamic range of HDRx™ is actually all available in the same frame (maybe not exactly simultaneously, see "magic motion", but in the same frame).

Ok... perhaps I shouldn't have complicated the question by mentioning HDRx.

If the Brightness Adoption Level of human sight is 6.5 stops at any one time. What is it for digital camera's like the RED one or the Epic etc?

 

[Hrvoje Simic]

The eye actually adjusts its sensitivity to different lighting conditions (chemical changes in the eye and adjusting the iris) and can only see 6.5 Stops simultaneously.

What "eye sees simultaneously" and what is final vision are two different things. Picture created in the mind is constantly refreshed and assembled as a higher DR image. Also, what "eye sees" many times is not even perceived consciously because subconscious systems define it as "redundant" and don't pass it into consciousness to avoid the overflow of data.

Don't take every incomplete "scientific" thesis as a fact of life because science fails where "exact" and "quantifiable" end. Also, humans tend to evolve upon realization of own limits.

Anyone who will shoot Epic in HDR mode will get a good impression of real visible DR and my guess is that it will vary between 14-16 stop range from person to person.

 

[Gavin Greenwalt]

It suggests the human eye has the Dynamic Range of 20 Stops but this is *not* available simulatenously. The eye actually adjusts its sensitivity to different lighting conditions (chemical changes in the eye and adjusting the iris) and can only see 6.5 Stops simulataneously. This range is known as the Brightness Adoption Level.

Unless I have super-vision I would say that's almost certainly 100% untrue. I'm looking out my window right now at a light colored building in direct sun and can see detail on a black chair indoors in a corner in shadow. There's at least 11-12 stops of simultaneous dynamic range there.

 

[Jody Neckles]

Unless I have super-vision I would say that's almost certainly 100% untrue. I'm looking out my window right now at a light colored building in direct sun and can see detail on a black chair indoors in a corner in shadow. There's at least 11-12 stops of simultaneous dynamic range there.

Indeed, I really don't get what this artical is saying about the human eye only have 6.5 stops at any one time.....

 

[Cail Young]

Unless I have super-vision I would say that's almost certainly 100% untrue. I'm looking out my window right now at a light colored building in direct sun and can see detail on a black chair indoors in a corner in shadow. There's at least 11-12 stops of simultaneous dynamic range there.

Can you see them both without moving your eyes? If not... then you are adapting, although you may not be consciously aware of it.

 

[Andy Jarosz]

Can you see them both without moving your eyes? If not... then you are adapting, although you may not be consciously aware of it.

Right. You're brain is like the greatest, quickest automatic camera in existence. Every time you change where you're looking, it's changing everything else to compensate--focus, iris, dynamic range. If you think about the size of the fovea, you don't need any more then 6.5 stops at a time.

 

[Gavin Greenwalt]

Can you see them both without moving your eyes? If not... then you are adapting, although you may not be consciously aware of it.

Yes. They're right next to each other and I'm doing my best to focus on something between the two haha. It's tricky I admit.

I also know the lighting at night in my apartment pretty well from photographing in here and I know the 5D blows out a couple spot light highlights when my black leather chair is exposed above black. Even in my peripheral it's easy to tell those points aren't blown out. I would say the eye has at least as much sensitivity at any given time as a 5D.

 

[Jody Neckles]

What "eye sees simultaneously" and what is final vision are two different things. Picture created in the mind is constantly refreshed and assembled as a higher DR image. Also, what "eye sees" many times is not even perceived consciously because subconscious systems define it as "redundant" and don't pass it into consciousness to avid the overflow of data. .

when one considers what our eyes are actually doing without conscious thought..... it is an insanely advanced HDR system. I wonder what sort of tests could prove we only see 6.5 stops in one instant????

I've just followed the author JD Vandenberghe on Twitter @jdvdb.... perhaps we can get him to join the discussion on REDUSER

 

[Gavin Greenwalt]

This paper says at least 5 orders of immediate simultaneous dynamic range. 8 with adaptation.

The Eye and High-Dynamic-Range Vision

5 orders of magnitude which would 13.28 stops (log base 2 of 10,000)
and
8 orders of magnitude which would be 23.25 stops (log base 2 of 10,000,000)

Perhaps the author mistook scientific "orders of magnitude" for "stops".

 

[Craig Ryan]

That sounds accurate, Gavin.

I've done tests with a spot meter, and I recall it was around 13 or 14. Maybe 15.

It's really hard to judge if you're actually seeing detail, or if your visual memory is filling in the gaps of something that's familiar to you (a black chair in your room for example). The brain and eye are WAY more complex than a camera sensor/film obviously, so I don't think we're going to get very far comparing the two.

 

[David Mullen ASC]

I think this just shows the problem of thinking that the "ideal" cinema process will recreate human vision -- the brain is constantly adjusting levels, focusing on details, shifting concentration, and even EDITING the image. Walter Murch in "The Blink of the Eye" talks about this phenomenon -- if your eyes scan a room from left to right, taking in the detail for the first time, you tend not to blink, and then if you repeat the move, you tend to blink in the middle of the move, basically "cutting to the chase" because you've already seen the room once, your brain wants to move on to something new, so it cuts out the boring bits by blinking, i.e. relaxing concentration.

So watching movies is a different activity than watching reality, and all the better for it, I can watch reality for free every day!

What matters is perception, a constructed reality that "feels" real even if it actually highly artificial.

 

[Thor Melsted]

We also need to keep in mind (as Craig Ryan and David Mullen just pointed out) what our brain is doing to fill in blanks in our vision at any given time. The number of rods and cones in our eyes simply doesn't add up to the "resolution" our eyes are capable of seeing at any given moment. A lot of it is our brain filling in blanks from memory. E.g. we only see color straight on because the 6 or so million color sensitive cones are concentrated in the middle surrounded by about 120 million light sensitive rods. So if you see something you have never seen before enter your peripheral vision, you won't know what color it is until you look directly at it. However, from that point forward, the brain knows, so you'll continue to see its color even in your peripheral. Our vision is based on continuous scanning, to absorb more information. Which is why we notice motion very easily from the corner of our eye but detecting color changes without looking straight at it is next to impossible.
The dynamic range of human vision far exceeds what the eye itself is capable of doing for the exact same reason. Our brains fill in the blanks and allow us to retain detail in areas in which our eyes are no longer able to detect detail due to either lack of light or an overabundance of it, but it does this so well we're impervious to it.

 

[Gavin Greenwalt]

Ok I'm calling total BS on that author. He's plagiarizing a book which means Automatic Fail. And the actual book's graphs show between 10-14 stops of dynamic range in experimentation. So his 6.5 number seems to be pulled out of his ass and most of the rest is just plagiarized.

As to "memory color.." Ummmmm no I'm going to have to call shenanigans on that one as well. It's easy to test. I just loaded up my screen with a solid color layer and am clicking around in the color picker. It's not very accurate but I can tell "redish" "blueish" "greenish" "purplish" even in my extreme peripheral.

 

[Thor Melsted]

As to "memory color.." Ummmmm no I'm going to have to call shenanigans on that one as well. It's easy to test. I just loaded up my screen with a solid color layer and am clicking around in the color picker. It's not very accurate but I can tell "redish" "blueish" "greenish" "purplish" even in my extreme peripheral.

Then your brain has adapted to the shades of gray associated with those colors, or there are enough cones getting photons from those shades to trigger a color recognition (they're concentrated in the center, but they do exist in other parts of the eye).

 

[Gavin Greenwalt]

Then your brain has adapted to the shades of gray associated with those colors, or there are enough cones getting photons from those shades to trigger a color recognition (they're concentrated in the center, but they do exist in other parts of the eye).

I guess it depends on what you consider extreme peripheral. Considering, I believe, we have a little less than 90* outward fov I would say 75* is pretty extreme.

 

[Thor Melsted]

Yes, 75 degrees is pretty good. It's not a narrow 20 degree cone or anything. I never said it was. But note that this is only for red and yellow. Green is at 60 degrees, which is a pretty small cone.

OK, try this. Look straight forward. Focus on your webcam or something straight ahead. Now hold your hands up to about behind your ears, but about a foot to the side from your head. Wiggle your fingers. Then move them forward until you can detect the motion of your fingers. That's the extreme of your peripheral vision. Between there and ~75 degrees is the zone you'll have little ability to detect colors, if any.

It's actually pretty interesting that green has a narrower fov than blue, considering the density of the cones sensitive to each color. Red at 64%, green at 32% and blue at 2%.

So, my example is probably rather exaggerated, but the point remains that we do not see everything at once - we're continually scanning and combining visual stimuli to build a whole picture, in a completely seamless manner.

 

[Jody Neckles]

I think this just shows the problem of thinking that the "ideal" cinema process will recreate human vision

I'm not a photographer for a living but my dream camera would simply capture an image, in terms of dynamic range, as I see it.

The 'barn' shot resonates with that dream.

As you suggest the best way to achieve is probably not to mimic the way the human eye (and brain) works.... that might be a little tricky