3

I am confused with the science of electronics, regarding cameras and sensors. I have read the posts and answers here and elsewhere. All I walk away with is the electronic specs of cameras and sensors.

I can relate to signal to noise ratio being relative to detail in a sensor. That can be affected by heat, design and age. But the ground level facts are that our cameras capture detail and tonal values. Any other features are just shiny distractions from the basic function.

I accept the fact that my Canon 5D Mk2 has less detail than the 5D Mk3. I accept that my camera has more noise than the 5D Mk3. But both cameras are within a half stop on tonal range. DxO labels my camera as having a 9-stop tonal range. My field tests agree with that finding. Yet it is repeated way too often that digital cameras have more DR than we need for a high dynamic range scene.

What dynamic range is being referenced here? Total? Tonal DR? Total tonal shift DR? Total detail DR? Total signal to noise DR? or what?

Am I wrong that we humans can see a dynamic tonal range of 25 stops? As well, depending on the author of article, book or video, we have roughly a potential of a 25-27 stops of tonal range in landscape scenes. Can someone boil this down to real world reality?

I practice a modified version of the digital zone system and my real world results seem to be very different than most specs hype. My field test of my Canon 5D Mk2 returned the following values: 9 stops of total tonal range, 7 stops of total tonal shift range, and 5 stops of total detail range. Respectfully curious!

inkista
  • 52,065
  • 10
  • 88
  • 161
backountry
  • 31
  • 2
  • 7
    "Yet it is repeated way too often that digital cameras have more DR than we need for a high dynamic range scene." I think this is the first I've heard this. I also am not aware that HDR has a specific minimum number of stops required. Sources? – Dan Wolfgang Dec 09 '15 at 18:09
  • 1
    The use of 'high dynamic range' really refers to tonemapping. It would be impossible for a camera to catch a genuinely HDR source since the whole point of the process is to use multiple exposures to create a composite source with a larger dynamic range than could be captured in one frame (since that would be a 'normal' dynamic range.) Therefore, no matter how many times the opposite might be repeated, it will always be untrue. – James Snell Jan 09 '16 at 18:35

1 Answers1

2

I am not familiar with the terms of 'tonal shift range' and 'detail range'. If I understand your question correctly I'd like to adress the dynamic range part of it however.
First of all:

Dynamic range, abbreviated DR or DNR,[1] is the ratio between the largest and smallest values of a changeable quantity, such as in signals like sound and light. - https://en.wikipedia.org/wiki/Dynamic_range

Now for the eye we are talking about a combination of eye and brain which basically does aperture bracketing automatically all the time, while the brain creates an HDR image of the scene for us. Wikipedia references 10-14 stops for that, however with longer adaption time (think of going from a brightly lit room outside directly and after 30 minutes of waiting outside), your value gets more correct (It should be something in the twenties). This very nice thread on our photo-stackexchange covers more than I am capable to say from the top of my head: How does the dynamic range of the human eye compare to that of digital cameras?

The value that is referenced when one speaks of

that digital cameras have more DR than we need for a high dynamic range scene

is the dynamic range (nothing fancy to add). You have to differenciate however of the DR your camera is capturing, and the intensity variation which the scene displays. If you photograph a forest against the sun, the DR of the scene is incredibly high and your sensor will not capture brightness differences in the shadows and/or in the highlights at some points of the image. If you take an image of a white bedsheed that is equally lit, the dynamic range can be under a stop and your sensor will not have any problems capturing the brightest and the darkest part of the scene.

The following question should be: which dynamic range does a typical "high dynamic range scene" have? In my opinion a highly philosophical question, also depending on what your typical subjects are. For landscapes, I would regard this to be an evening, close-to-sundown situation as often photographed, and just by a guess I would say that the DR in the scene will easily overcome 15 stops, which a modern sensor cannot cover.

Community
  • 1
kamuro
  • 1,060
  • 5
  • 18
  • I have no problems with your reply. For a landscape photographer, high dynamic range does include....sun, brilliant reflection the sun off of water, etc. Not all landscapes present themselves at the golden hour.DR in the upper 20's do exist in the real world of photography. – backountry Dec 10 '15 at 14:31
  • As for the term Dynamic Range..... – backountry Dec 10 '15 at 14:36
  • DXO uses DR as the signal to noise ratio of a sensor. If sensor improvement or software improves on a cleaner signal, then greater detail can be achieved. If that is correct, then there is total understanding. The DR that DXO states is: relative to the noise that the sensor creates. That noise can be affected by, sensor heat, crowding of diodes, good or poor algorithms. Totally understandable. Noise is a feature of the tool and not a feature of image capture. Signal to noise is a lab measure. We still capture Only Detail, and Tones with our camera. DXO has addressed detail noise. – backountry Dec 10 '15 at 15:08
  • So their DR could be considered as the Signal to Noise Dynamic Range. It is stated as a maximum value. It is not necessarily the true DR of your camera, due to noise variables and manufacturer standards. Where does signal to noise fit into the exposure equation? – backountry Dec 10 '15 at 15:14
  • The signal to noise ratio is determined by the noise in reference to the maximum possible signal, which is a digital value determined by the bits that are recorded in the camera from the sensor, IF the sensor can really put out a measure for incident light that is distinguishable between the digits (I could also express 256 values in a system of 512, but the difference between 373 and 374 will be nonexistent.) To my understanding, this means: how much light intensity dynamic range detail will you not be able to resolve when distiguishing brightness values. – kamuro Dec 10 '15 at 17:15
  • And it will cut a bit of the overall light intensity dynamic range, since you cannot use the lowest intensity values, as all information is lost in the noise. If this is accepted, strengthened or refused by other users, I'd gladly edit the information into the answer. – kamuro Dec 10 '15 at 17:17