Sign in with
Sign up | Sign in
Your question

Scene range vs dynamic range

Last response: in Digital Camera
Share
Anonymous
June 30, 2005 1:55:32 PM

Archived from groups: rec.photo.digital (More info?)

Apparently the term dynamic range lends itself to much misunderstanding.
So, I propose we try some new terminology so that the various factors
in evaluating image capture can be discussed.

What I think most photographers are interested in is the scene range.
That is the ratio of the brightest portion of a scene that can be
captured without clipping compared with the darkest region which can
be captured without "too much" noise (or grain).
The bright limit is fairly easy to determine with slide film and digital
sensors, it is the point where any more exposure does not produce any
change in the value stored in the image. For slide film it is clear
film, for digital it is the max readout (255 or whatever the scale is).

On the dark side things get a little more complicated, with no exposure
at all film will produce some density (usually inherent fog) and digital
will produce a non-zero value (various noise sources).
At this point the user or manufacturer needs to establish a criterion
for how much above this residual is necessary to consider the resulting
values as "real" information. For film it is usually some fixed density
increment like .2 above fog or whatever. For digital there seems to be
no totally agreed on standard, but there are several reasonable choices
around.

Now the difference between these two limits set the amount of scene
range that can be captured. For slide film this is usually around 6 or 7
stops, for black and white it is around 9-10 stops (the zone system, for
example). For color negative film it seems to be about 10+ stops.
For digital capture people quote anywhere from 6 - 14 stops. It is not
clear to me if these difference are due to the wide range of
capabilities of digital sensors or the different criteria used to
evaluate an image.

Now once we have established the scene range we have the other factor
which is how fine a gradation between values can be record. Film being
an analog medium the answer should theoretically be infinite, but in
most cases it is usually considered to be some small finite density
change like .02 or .03 on the film. For prints the value depends on the
acuity of the eye and is usually taken to be about 1/2% change in
brightness for medium ranges and greater for the darkest areas. So
people commonly state that a print can show about 50 or so discrete
shades.

For digital the issue gets confused by the software used to evaluate the
captured data. Notice the scene range discussion is independent of the
software. So if the camera software only delivers 8 bits then the scene
data will be divided into one of 256 possible values. More bits allows
for more steps, however the maximum and minimum values don't change.
When looked at this way the discussion of "dynamic range" is clarified.

The maximum brightness of a scene can be reported as 256 or 4096 or
100%, but this does not affect where the loss of detail occurred.
The number of bits has a use when trying to make adjustments to the
digital original after capture. If an image is represented by a small
number of bits and their values are shifted by a large amount then gaps
in the values can occur and the scene gets to look blocky or pixelated.

There are many discussions on how many bits are needed to avoid this and
how visible the effect is, but theoretically more bits allow more
extreme editing.
To use an analogy the scene is represented by a foot long ruler, the
darkest portion is at 0 and the brightest portion is at 12". The object
measured remains a foot long regardless of whether the markings on the
ruler are 1 inch apart or 1/16 inch apart. It is just that somewhere in
the middle of the ruler we can make a more accurate measurement with the
finer divisions. This is the difference between accuracy and precision.

Dynamic range is the correct scientific term for what is meant by the
"useable" ratio of the biggest to the smallest signal values whether
these are from light or sound or radio waves or whatever. However, for
a working photographer I think the concept of scene range might be less
confusing and more useful in practice.

--
Robert D Feinman
Landscapes, Cityscapes and Panoramic Photographs
http://robertdfeinman.com
mail: robert.feinman@gmail.com
Anonymous
July 4, 2005 12:58:22 PM

Archived from groups: rec.photo.digital (More info?)

Hello Robert

I have to say that this is one of the most interesting posts on this
subject I've read and I will print it out and read it again too.

I'm a bit surprised that it has gone unanswered, but then again
sometimes a more indepth approach won't get read fully and, as a
consequence, not grasped.

I've been to your site. You have great images there - a lot of work!

For DSLR/SLR Amateurs & Novices:
www.theimageplane.net
July 4, 2005 9:30:04 PM

Archived from groups: rec.photo.digital (More info?)

Robert Feinman wrote:
> Apparently the term dynamic range lends itself to much misunderstanding.
> So, I propose we try some new terminology so that the various factors
> in evaluating image capture can be discussed.
>
> What I think most photographers are interested in is the scene range.
> That is the ratio of the brightest portion of a scene that can be
> captured without clipping compared with the darkest region which can
> be captured without "too much" noise (or grain).
> The bright limit is fairly easy to determine with slide film and digital
> sensors, it is the point where any more exposure does not produce any
> change in the value stored in the image. For slide film it is clear
> film, for digital it is the max readout (255 or whatever the scale is).
>
> On the dark side things get a little more complicated, with no exposure
> at all film will produce some density (usually inherent fog) and digital
> will produce a non-zero value (various noise sources).
> At this point the user or manufacturer needs to establish a criterion
> for how much above this residual is necessary to consider the resulting
> values as "real" information. For film it is usually some fixed density
> increment like .2 above fog or whatever. For digital there seems to be
> no totally agreed on standard, but there are several reasonable choices
> around.
>
> Now the difference between these two limits set the amount of scene
> range that can be captured. For slide film this is usually around 6 or 7
> stops, for black and white it is around 9-10 stops (the zone system, for
> example). For color negative film it seems to be about 10+ stops.
> For digital capture people quote anywhere from 6 - 14 stops. It is not
> clear to me if these difference are due to the wide range of
> capabilities of digital sensors or the different criteria used to
> evaluate an image.
>
> Now once we have established the scene range we have the other factor
> which is how fine a gradation between values can be record. Film being
> an analog medium the answer should theoretically be infinite, but in
> most cases it is usually considered to be some small finite density
> change like .02 or .03 on the film. For prints the value depends on the
> acuity of the eye and is usually taken to be about 1/2% change in
> brightness for medium ranges and greater for the darkest areas. So
> people commonly state that a print can show about 50 or so discrete
> shades.
>
> For digital the issue gets confused by the software used to evaluate the
> captured data. Notice the scene range discussion is independent of the
> software. So if the camera software only delivers 8 bits then the scene
> data will be divided into one of 256 possible values. More bits allows
> for more steps, however the maximum and minimum values don't change.
> When looked at this way the discussion of "dynamic range" is clarified.
>
> The maximum brightness of a scene can be reported as 256 or 4096 or
> 100%, but this does not affect where the loss of detail occurred.
> The number of bits has a use when trying to make adjustments to the
> digital original after capture. If an image is represented by a small
> number of bits and their values are shifted by a large amount then gaps
> in the values can occur and the scene gets to look blocky or pixelated.
>
> There are many discussions on how many bits are needed to avoid this and
> how visible the effect is, but theoretically more bits allow more
> extreme editing.
> To use an analogy the scene is represented by a foot long ruler, the
> darkest portion is at 0 and the brightest portion is at 12". The object
> measured remains a foot long regardless of whether the markings on the
> ruler are 1 inch apart or 1/16 inch apart. It is just that somewhere in
> the middle of the ruler we can make a more accurate measurement with the
> finer divisions. This is the difference between accuracy and precision.
>
> Dynamic range is the correct scientific term for what is meant by the
> "useable" ratio of the biggest to the smallest signal values whether
> these are from light or sound or radio waves or whatever. However, for
> a working photographer I think the concept of scene range might be less
> confusing and more useful in practice.
>
My prediction is that when (if ever) your proposal takes effect, it will be widely
misunderstood. For one hting, the marketers at camera makers will want to bend the
definition to make their products look good.
!