Doug Kerr
Well-known member
"Dynamic range" has for many years been used to characterize the ability of a measuring instrument, recording system, signal processing amplifier, or the like to satisfactorily deal with, in the same "session", a wide range of "values" of the quantity of importance (which might be, for example, the instantaneous voltage of an audio waveform).
The modifier "dynamic" is meant to indicate what I spoke of simplistically as "in the same session". For example, an audio waveform recording system may be able to successfully capture two signals whose voltage differ by a ratio of 150 dB but only in the sense that we will change the sensitivity adjustment ("gain") of the recorder between trials. If we leave the gain at a fixed setting, then the system may only be able to successfully capture signals whose amplitudes have a ratio of, say, 60 dB.
Thus, we might colloquially interpret "dynamic" as meaning "on the fly".
A concern with photographic processes, analog or digital, is their ability to successfully capture (for a single "shot") a wide range of scene luminance, that is, to be able to properly capture the detail (which is manifest by local changes in luminance) of a scene in which there are areas of interest ("highlight" and "shadow" areas) where the general luminance differs by a great ratio.
Various measures of the ability of a digital camera to "successfully" capture detail over a range of general luminance have been devised (part of the trick is deciding what "successfully" should mean), and we are interested in the "score" of particular camera systems in this regard - what we call the "dynamic range" of the cameras.
As a complement to this, we may think of the range of base luminance in a scene as the "dynamic range" of the scene, where "dynamic" means "over this scene at a particular instant of time". Some would argue that the term should only be applied to the ability of a camera to capture a certain range of luminance, and not to the range of luminance exhibited by a scene. But both sides of that argument have merit, and my point here is not to pursue it.
Because today's generally-available cameras do not have a dynamic range sufficient to capture the luminance range of many scenes we wish to photograph, clever techniques have been contrived to overcome this limitation. A basic one is to take two or more images of the same scene with different photographic exposure. The lower exposure image may successfully capture the detail in the "highlight" areas, and the higher-exposure image the detail in the shadow area.
Processing software then combines these two or more images in a single image, recoded (in computer memory, for the time being) in an encoding system having sufficient dynamic range to capture the composite image, which we hope faithfully mirrors the luminance of the entire scene.
But now we encounter a problem. It may well be that the format in which we wish to deliver the composite image does not have sufficient dynamic range for the purpose. More importantly, it may be that the form in which the image will be presented to the viewer (display on a screen, or as a print) cannot reliably reproduce the range of luminance in the composite image.
So we have our processing software do some perceptual tricks for us. We arrange for the high-luminance ("highlight") regions of the image to be presented near the top limit of the relative luminance range of the display medium, with their detail intact. Similarly, the low-luminance ("shadow") regions are presented near the bottom limit of the relative luminance range of the display medium, with their detail intact.
While the delivered image does not actually have a luminance range any greater than it ordinarily does, we do have the perceptual benefit we mostly seek in high-dynamic range reproduction: the ability to see the detail in both highlight and shadow areas of the scene.
Now, if we describe this result using the term "high dynamic range", just what do we mean?
Do we mean that the scene had a higher range of luminance than we could have dealt with under normal technique? Probably.
Do we mean that our overall system has captured a wider range of luminance than otherwise? Yes, it has - inside the process.
Do we mean that the overall system has captured and delivered to us a wider range of luminance that otherwise? No, it hasn't.
Do we mean that our overall system has allowed the viewer to enjoy the detail of a "higher dynamic range" scene than we could otherwise? Yes it has.
There is no single recipe for optimally "remapping" the high dynamic range composite image into a lower dynamic range delivered image. Thus the software packages we use for this work have elaborate provisions for controlling their behavior.
As with any new tool, people find ways to use it for things other than its original purpose. Thus photographic workers have found that by processing a single taken image in such software, and making various adjustments to its controls, they can transform the image so it has quite different appearance than we would usually have. And they present the result of so doing as a piece of photographic art. And it is, just as are "monochrome", or "sepia", or "duotone", or "posterized", or "pixellated" delivered images.
My point here is not to apply any arbitrary critique of such work as a genre. As with all art, each piece needs to be considered on its own merits, under whatever scheme of criteria we care to adopt.
But please, when you do that kind of work, don't label it "high dynamic range". If you do, I may be compelled to ask, "Just what dynamic range does that image have, anyway? Or maybe, what dynamic range did the original scene have?"
Best regards,
Doug
The modifier "dynamic" is meant to indicate what I spoke of simplistically as "in the same session". For example, an audio waveform recording system may be able to successfully capture two signals whose voltage differ by a ratio of 150 dB but only in the sense that we will change the sensitivity adjustment ("gain") of the recorder between trials. If we leave the gain at a fixed setting, then the system may only be able to successfully capture signals whose amplitudes have a ratio of, say, 60 dB.
Thus, we might colloquially interpret "dynamic" as meaning "on the fly".
A concern with photographic processes, analog or digital, is their ability to successfully capture (for a single "shot") a wide range of scene luminance, that is, to be able to properly capture the detail (which is manifest by local changes in luminance) of a scene in which there are areas of interest ("highlight" and "shadow" areas) where the general luminance differs by a great ratio.
Various measures of the ability of a digital camera to "successfully" capture detail over a range of general luminance have been devised (part of the trick is deciding what "successfully" should mean), and we are interested in the "score" of particular camera systems in this regard - what we call the "dynamic range" of the cameras.
As a complement to this, we may think of the range of base luminance in a scene as the "dynamic range" of the scene, where "dynamic" means "over this scene at a particular instant of time". Some would argue that the term should only be applied to the ability of a camera to capture a certain range of luminance, and not to the range of luminance exhibited by a scene. But both sides of that argument have merit, and my point here is not to pursue it.
Because today's generally-available cameras do not have a dynamic range sufficient to capture the luminance range of many scenes we wish to photograph, clever techniques have been contrived to overcome this limitation. A basic one is to take two or more images of the same scene with different photographic exposure. The lower exposure image may successfully capture the detail in the "highlight" areas, and the higher-exposure image the detail in the shadow area.
Processing software then combines these two or more images in a single image, recoded (in computer memory, for the time being) in an encoding system having sufficient dynamic range to capture the composite image, which we hope faithfully mirrors the luminance of the entire scene.
But now we encounter a problem. It may well be that the format in which we wish to deliver the composite image does not have sufficient dynamic range for the purpose. More importantly, it may be that the form in which the image will be presented to the viewer (display on a screen, or as a print) cannot reliably reproduce the range of luminance in the composite image.
So we have our processing software do some perceptual tricks for us. We arrange for the high-luminance ("highlight") regions of the image to be presented near the top limit of the relative luminance range of the display medium, with their detail intact. Similarly, the low-luminance ("shadow") regions are presented near the bottom limit of the relative luminance range of the display medium, with their detail intact.
While the delivered image does not actually have a luminance range any greater than it ordinarily does, we do have the perceptual benefit we mostly seek in high-dynamic range reproduction: the ability to see the detail in both highlight and shadow areas of the scene.
Now, if we describe this result using the term "high dynamic range", just what do we mean?
Do we mean that the scene had a higher range of luminance than we could have dealt with under normal technique? Probably.
Do we mean that our overall system has captured a wider range of luminance than otherwise? Yes, it has - inside the process.
Do we mean that the overall system has captured and delivered to us a wider range of luminance that otherwise? No, it hasn't.
Do we mean that our overall system has allowed the viewer to enjoy the detail of a "higher dynamic range" scene than we could otherwise? Yes it has.
There is no single recipe for optimally "remapping" the high dynamic range composite image into a lower dynamic range delivered image. Thus the software packages we use for this work have elaborate provisions for controlling their behavior.
As with any new tool, people find ways to use it for things other than its original purpose. Thus photographic workers have found that by processing a single taken image in such software, and making various adjustments to its controls, they can transform the image so it has quite different appearance than we would usually have. And they present the result of so doing as a piece of photographic art. And it is, just as are "monochrome", or "sepia", or "duotone", or "posterized", or "pixellated" delivered images.
My point here is not to apply any arbitrary critique of such work as a genre. As with all art, each piece needs to be considered on its own merits, under whatever scheme of criteria we care to adopt.
But please, when you do that kind of work, don't label it "high dynamic range". If you do, I may be compelled to ask, "Just what dynamic range does that image have, anyway? Or maybe, what dynamic range did the original scene have?"
Best regards,
Doug