Dean M. Chriss
Photography

Dynamic Range in Perspective
To see or not to see highlight and shadow details, that is the question.
February 11, 2019

Color Coordinated Kestrel
Color Coordinated Kestrel

Photo forum pontificators insist we need gobs of DR, but when I look at my ancient Kodachromes, which I assumed had extremely limited dynamic range, they look alright. How much DR do they really have? How much DR can scenes in nature have? How much DR can human vision accommodate? How much DR do I need? How is all this stuff related? I set out to put the topic of dynamic range (DR) into perspective for myself.I know something about physics but not physiology, so I spent lots of time researching this stuff and reconciling differing information, much of it wrong, from multiple sources. Since I did the work I thought it may as well be put it out here so other curious organisms don't need to repeat the exercise. Condensing my thoughts and re-looking-up some blurred details took more time, and this is as good as I am willing to make it. You're welcome! 

Dynamic range (DR) and contrast ratio are both measures of the difference between the luminance of the brightest and darkest objects that are simultaneously distinguishable. Pure white without any detail lies at and above the bright end of the range, while pure black without any detail lies at and below the dark end of the range. Everything between the two will render detail. Dynamic range is usually expressed in f-stops. Contrast ratio is expressed as a ratio between the luminosity of the brightest and darkest objects that are simultaneously visible. Because each f-stop represents a doubling of light, the equivalence between the two is given by the equation
Contrast Ratio = 2^(f-stops), as shown in the following table.

F-stops Contrast Ratio
6 64:1
8 256:1
10 1024:1
12 4096:1
14 16384:1
16 65536:1

Humans

The range between the most intense light the human eye can safely handle and the darkest things we can see in the darkest environments is almost 30 stops, but there's a big catch. We cannot see both ends of the scale simultaneously, so this is not dynamic range in any useful sense. In fact it takes between 30 minutes and several hours for our vision to adapt between the extremes. It's even more complicated than that. Human vision has three ranges, scotopic, mesopic, and photopic.

Scotopic vision occurs in dark environments with luminance less than 0.001 lux, when only the rods in the retina are responsible for vision. Rods are not sensitive to color, so vision in the scotopic range is monochromatic. Starlight, (outdoors away from artificial light sources on a clear night with no moon) produces a luminance of about 0.001 lux. The scotopic range is where our eyes have their greatest dynamic range; about 20 stops for a contrast ratio of about 1,000,000:1. That's enormous, but it is monochromatic.

Mesopic vision occurs in environments with luminance between 0.001 and 10 lux. Both the rods and cones in the retina are used for vision in this range. The proportion of rods to cones depends on the luminance, as does the available dynamic range, which varies from about 20 stops in monochrome on the dark end to about 10 stops in color on the bright end.

Photopic vision occurs in environments with luminance greater than 10 lux. Only the cones are used for vision in this range, and this is where our best temporal, image, and color resolution occurs. It is also the range in which we spend most of our time. Common luminance levels range from around 20 lux in public areas with dark surroundings (like the darker areas in a public parking garage at night) to 100 lux in elevators, stairwells, and storage spaces to 150 lux at home to 300 lux in classrooms to about 750 lux in a supermarket. In the photopic range we can capture a dynamic range of about 10 stops or a contrast ratio of about 1024:1. Only about 6.5 stops of this dynamic range is available instantaneously, but the visual system dynamically adjusts its sensitivity (ISO) to optimally place the available dynamic range according to the brightness of the specific area the person is trying to see. The iris (aperture) also adjusts dynamically and has an adjustment range a bit less than 3 stops. Our brains integrate what we see over brief periods into what could be called the HDR image we perceive. In total we end up with about 10 stops of usable dynamic range.

Scenes in real life can have a dynamic range far greater than 30 stops, and far greater than the 10 stops we can see in full color.

Photographic Output Media

In spite of marketing claims, when measured with good calibration equipment, the best LCD monitors have a dynamic range of barely 10 stops. That's a contrast ratio of 1024:1, which is the same as that of our eyes. We have to remember that this is what a calibration tool "sees", and it excludes all ambient light from the area it measures. Humans do not normally do that. In normal viewing conditions the dynamic range is more like 9, or even 8, stops. Photographic prints have a dynamic range of about 6.5 stops, assuming good lighting with no glare from the printed surface and no glazing over the print.

Cameras

Back in the days of film, Kodachrome slides had a dynamic range of about 6 stops, or a contrast ratio of about 64:1 after accounting for the noise floor (grain). Kodak Gold 200 color negative film was the best with about 7 stops of dynamic range, while Fuji Velvia had about 5 stops. Canon's 2004 vintage EOS 10D digital camera had dynamic range of about 8 stops. Most current digital cameras equal or exceed 10 stops of dynamic range while the best medium format models can reach 13 stops. That is the same or more than human vision or any display medium, but far less than can sometimes be encountered in nature.

When capturing an image we adjust camera exposure settings to optimally place its available dynamic range according to the brightness of the specific area we are trying to photograph. Sound familiar? If the scene has more dynamic range than the camera we must either choose which end of the scene's range to clip (let go to black or white without detail), or take bracketed exposures to later combine and produce an HDR (High Dynamic Range) image. This capture process is mapping the larger dynamic range of a scene into the smaller dynamic range of the camera. After an image is captured a large part of optimizing it involves mapping the dynamic range of the generated image file(s) into the still smaller dynamic range of a monitor or print. The fact that so much more dynamic range is available in the HDR files than we can display, print, or even see, is one reason "overcooked" HDR photographs look so odd. They show more detail in shadows and highlights than would be visible if we were looking at the original scene and the contrast necessarily goes flat. High contrast scenes in human vision often have clipped blacks and/or whites at any given instant, depending on where we look. Because our eyes move rapidly, zero in on different parts of a scene, and constantly adjust, we tend to not notice.

Cameras with extended dynamic range allow photographers to place the available dynamic range of the camera less accurately at the time images are captured and still produce good results. A wide dynamic range can be captured and later mapped to the lower dynamic range of the display media more carefully in more comfortable surroundings. There is also no real issue with showing more highlight and shadow detail in a photograph than could be seen in reality. Taken beyond some subjective point the results begin to look unreal (because they are), but artistic license allows us to make unreal looking photographs!

Conclusion

There are no rules but generally speaking it is not necessary to have details in the smallest and deepest shadows or brightest highlights of a high contrast scene. For instance, specular highlights on water or shiny objects ought to clip a little because they are way too bright to see in person. High contrast scenes often look unnatural when we can see too much detail in the deepest and darkest recesses. How good or bad any of this looks is an artistic question that depends on context, intent, and how large those black and white areas without detail are. And there are dramatic exceptions. For instance, a properly exposed subject on a pure black background without detail can look wonderful. Similarly, bright backgrounds that have little or no detail can often go well beyond the clipping point (pure white) with good effect. There's not much point in having an underexposed image of a bird against an unclipped sky that had no detail anyway, and a nice subject against a pure white background can be striking.

Looking at all of the numbers we see that people can take in about 10 stops of dynamic range. The highest dynamic range output medium is a good LCD monitor, which had an absolute maximum of about 10 stops of dynamic range, but more typically 8 or 9 stops. Digital cameras have between 10 and 13 stops of DR, and scenes in nature can exceed 30 stops of DR by a large margin. I'll leave any further conclusions to you.

Happy contrasting!

Dean