Digital BW, The Print

----- Original Message -----
From: Austin Franklin
To: DigitalBlackandWhiteThePrint@yahoogroups.com
Sent: Wednesday, March 27, 2002 8:51 AM
Subject: RE: [Digital BW] Dynamic Range Definitions and Print Tones


Hi Martin,

>> Dynamic Range, Some Definitions:

>The definition/concept of dynamic range is taught in any good EE program,
>and is very basic to most electrical engineers (NOT computer engineers,
they
>are not the same).  I have probably measured the dynamic range, and written
>the specs for, over 100 different systems.  I am an EE, and, as you know,
>have been designing both audio and video equipment for over 25 years.  I
>don't mean to sound smug, but it is unquestionable that my
>definition/understanding of dynamic range is correct.  If it was not, since
>I do this for a living, and am subject to direct peer review by hundreds of
>different engineers over my career, as well as thousands of customers and
>people who read my specs...and if I had it wrong, someone would have
pointed
>it out by now...and no one has.

>Most of the definitions you cited are "pedestrian" definitions written by
>non-technical people who are, for the most part, mis-explaining a concept
>that they apparently do not understand.  What you are calling "dynamic
>range" is merely static range.  Measuring the max and min value of
something
>merely gives you the STATIC range, NOT the DYNAMIC range.  Comparing them
by
>division or subtraction still give you a STATIC result, not a DYNAMIC
>result.

I guess they probably have a shortage of technical people at Cornell and
NASA to write up these definitions. <G>

You ask me to believe that all these other sources are wrong and you are
right. I don't mean to offend but I just can't do that. As I have said
before, you seem to be taking mathematical concepts from EE and assuming
that they directly apply to non-EE systems.

>> So what we have is a variety of definitions that seem to vary
>> depending upon what field you are in

>They vary simply because the people who wrote the incorrect definitions did
>not understand what it was they were saying.  Dynamic range IS dynamic
>range, no matter what the application.  The X-Rite definition you supplied
>is the best one as far as I am concerned.

All these people are wrong and you are correct? Austin, please. The X-Rite
one is good for a measuring instrument.

>> but the common element is
>> that they are looking for a meaningful way to describe a
>> relationship between the minimum and maximum as a ratio.

>That's the issue, they do not describe what they mean by minimum and
>maximum...  They are ambiguous terms.  These terms can mean either
amplitude
>or static values.  Per the diagram I posted in another post, they clearly
>mean amplitude.

The nature and the determination of the min and max will vary with what
physical property is under examination. The words "minimum" and "maximum"
are perfectly clear.

>> Some people say the dynamic range is the difference between the
>> min and max and some say it is the ratio.

>It is always a ratio, as defined by any dynamic range equation.

Austin, I cannot believe that you do not understand the difference of the
log of the max and the log of the min is the ratio of the max and min.

>  Forget web
>and dictionary definitions,

I should throw out the dictionary definitions? I don't see any reason to do
that. If that is what is required to prove your point you would seem to be
on thin ice.

> it's the equation that matters here, and only
>the equation.  The equation if finite and definitive.  Your understanding
>does not follow the equation.

>The dynamic range equation from "Digital Signal Processing in VLSI" by
>Richard J. Higgins is:

>Dynamic Range (dB) = log10 (largest signal/smallest discernable signal)

If I say Dynamic Range (dB) = log10(largest meaningful signal) -
log10(smallest meaningful signal) this is mathematically the same and if you
plug in the numbers you get the same answer.

>The diagram I posted previously is from this very book/equation.  It
clearly
>shows that largest and smallest are amplitude.  Also note DISCERNABLE.  To
>discern something means it must change with respect to what it is you are
>trying to discern it from, which in this case is it self.  You need two
data
>points for "discernability".  You can not discern something that is static
>(shows no change).

The criteria for determining or discerning the min and max values will vary
from field to field and may be different at opposite ends of the scale. In
audio the minimum is determined in relation to the background noise floor
but the maximum is chosen based on a certain degree of distortion. I am sure
what you say and your diagram are excellent relationships to use in
evaluating electrical systems but are not a necessarily a good way to
determine the min and max in all systems. For instance "noise floor" and
"third harmonic distortion" would not be helpful in deciding what the usable
min and max tones in a photographic print.

It is more important to talk about meaningful values rather than simply
discernable values. You have to look at the system and decide what is of
interest and produces usable information. This may be different from what is
discernable.

>> Some things such as
>> sound volume, CCD response and image density are not direct
>> measurements of a physical property but are calculated values
>> from actual properties given in logarithmic form.

>Absolutely not true.  You absolutely can measure the dynamic range of audio
>systems and CCDs and image density.  You CAN run an FFT on the data to
>arrive at the dynamic range, but you still have to do direct measurements
to
>arrive at the values you give to the FFT...you don't just make them up!

I am sorry but dynamic range is not measured, it is a calculated value.
Didn't you just give us an equation to do this? dB in audio are calculated
values derived from a direct measurement of pressure. This is where you are
missing my point. Some values are so common we have lost sight of the fact
that they do no represent actual physical properties. Print density is not a
physical property. It is a value calculated from a measurement of light flux
(a real physical property) by an instrument which does an internal
calculation and gives you a calculated density value.

In order for us to make a decision about the significance of the min and max
values we need to understand how they were derived.

>> So if a print has a white paper base reflectance of 95% (95% of
>> the light shining on this spot bounces off toward your eye) and
>> the deepest black has a reflectance of 4% (only 4% of the light
>> falling on this dark spot bounces off and the rest is absorbed or
>> scattered away from your eye.) The dynamic range would be the
>> maximum value of 95% divided by the minimum value of 4%, which is
>> 23.75.

>Because density is represented in a log form, and dynamic range is
>represented in a log form does not make them both the same.  Dynamic range
>and density range are NOT the same.

They are because of the math. The difference between two log values is a
ratio of the root numbers.

> How do you know you really measured 95% and not 95.876593254?

I did not measure anything. I asked that you assume actual values not
measurements. There is a difference between the real value and the
instrument readings.

>> Usually in photography we do not talk about reflectance but
>> instead use density. So for this case the paper base would have a
>> density of 0.0223 (the base 10 log of 4%)and the darkest black
>> 1.3979 (the base 10 log of 95%). Since these are log values
>> calculated from the reflectance you find the dynamic range by
>> subtracting them. 1.3979 minus 0.0223 is 1.3757 the dynamic range
>> and the range of the Density.

>You have NOT measured any dynamic range, as you have NOT measured the noise
>in the system.  What you have is the DENSITY range NOT the DYNAMIC range.
>Again, you can NOT relate dynamic range to density range simply because
they
>are expressed in log form.

Once again you make up your own definitions and ignore the published ones.
Noise is not required to calculate the dynamic range, only the meaningful
min and max.

You have repeatedly refused to answer my question as to what you think the
noise is in a photographic print.

>> So if you are talking about image density (a log value) the
>> dynamic range and the range are the same.

>No, absolutely wrong.  Because two things use the same numeric
>representation it does not mean they ARE the same.  Dynamic range is
>measured in DECIBELS, density is measured NOT in DB, but on a calibrated
>relative density scale.  NO density measurement equation will contain DB.
>That fact alone negates your entire premise that they are the same.

Density is calculated not measured. A densitometer measures the light flux
reflecting from or passing through the material. It then takes the
difference between this flux and its standard flux. It then converts the
flux to a percentage, divides it into 1, takes the log10 of the result and
displays it.

>OK, here is the Density measurement equation from "Introduction to
>Densitometry" published by the Graphics Communication Association:

>Density = log10 1/R (where R is the reflectance)

I believe I have stated that several times now.

>and to quote the book:

>"This is a pure definition, which means that it defines the relationship
>between light reflectance and density.  This definition numerically
>interprets density, and, more importantly, describes density in a manner
>that approximates the way in which the human eye sees objects."

Yes because our eyes do not respond to light in a linear fashion. Density is
a convenient way to look at reflectance but it is a calculated value not a
direct measurement.

>No where in this entire book do they talk about dynamic range.  Also,
>density measurements REQUIRE they be compared to a known calibrated sample,
>which is why you have to calibrate densitometers.  Measuring dynamic range
>does NOT require it be calibrated to any known sample, it is simply
relative
>in and of it self.  Density measurements aren't, they are relative to an
>absolute measurement...hence, they are STATIC.

Of course density is static. (Unless your prints are fading of course.<G>)
You would then seem to be saying that a photographic print does not have a
dynamic range.

>> Noise generally does not seem to be used in the calculation of
>> the dynamic range

>Absolutely wrong.  The "smallest discernable signal" in any analog systems
>IS the noise in the system.  In a digital system, as in number of bits
let's
>say, the minimum discernable signal is 1, so 4 bits has a dynamic range of
>log10(16/1) or 1.2dB.  The key is "smallest DISCERNABLE" signal, what ever
>that might be in that particular system.

Austin, this is just a semantic game. It just so happens that the smallest
meaningful signal in an analog system is equal to the noise floor. Again
what is the noise in a static photographic print and how does it relate to
either reflectance or density?

>> The best example would be in audio
>> systems where there is a background hum or hiss. This is what is
>> known as the noise floor and in audio the weakest or minimum
>> value is this noise level since you cannot hear any sounds that
>> have a volume lower than the noise.

>Again, you are using ambiguous terms here (weakest/minimum).  The weakest
>value is NOT necessarily the noise floor.  You can turn the volume down,
and
>decrease the noise to almost nothing, and that also doesn't take into
>consideration the noise before the volume control.  Also, the noise will be
>different depending on the volume, as when you add gain to a system, you
>also amplify the noise.

The entire idea, as I clearly stated, is to use the smallest meaningful
value which is at the noise floor in this example. I see no ambiguity in
that.

>> There is noise in the process of making a print since information
>> is lost or degraded as you move from scene to camera to print,
>> but once the print is finished the concept of noise does not
>> appear to apply

>Of course noise applies.  Anything that as distorted the actual image
>information has caused noise, anywhere in the system.  It is not relevant
if
>the print is "finished" or not.

Noise is only relevant in describing the output to the input in this case.
Once we are left with just the output hanging on the wall what is the noise?

>> Finally the "number of tones" present in a print does not depend
>> upon the dynamic range or the range

>Absolutely wrong...as I've said time and time again, and for some reasons
>you don't understand my examples.  The number of tones is SOLELY based on
>the dynamic range.  It's the definition OF dynamic range (as defined by any
>competent source).

The number of tones is only related to the dynamic range in digital systems
which have finite steps. Analog systems are stepless and do not have a
finite number of tones.

I chose my sources and I stick with them.

>How many tones does 8 bits represent?  256.  How many tones does 9 bits
>represent? 512.  Now, you can represent ANY density range with only two
>bits, as I have explained in another post.  You can represent any density
>range with ANY number of bits.  It is merely a representation.  BUT the
>dynamic range will be entirely different, but the density range will be the
>same.

 This is true in digital but does not relate to analog which has no bits.

>> A print
>> is an analog image and by definition has continuous tones that
>> flow from one to another without any step change or gaps.

>Absolutely wrong, again.  NO analog system is completely continuous.  There
>is NO such thing.  ALL systems have a resolution analog or not!  It's a
>fact.  Because you don't understand that it doesn't doesn't make it not
>true, it just means you don't understand/get that.

Austin, you choose not to acknowledge that the word continuous means
unbroken or that analog is defined as "a mechanism in which data is
represented by continuously variable physical quantities."

Once again if you want to reinvent the definition of words we have no basis
for a meaningful dialog.

>> Just as
>> the sun sets in a smooth motion and not in increments, so the
>> shades of gray flow in a continuous tone print.

>Your ability to SEE the sun setting absolutely is in increments.  It is
well
>knows that your vision has a resolution, and you can only discern things
>with the accuracy of so much of an arc.  It's a fact.

How I see it and what it does are hardly the same thing. As I have said
before, you keep wanting to equate the measurement of a property to the
property itself and this is not correct.

>Quoted from "Modern Optical Engineering" by Warren Smith:

>"The resolution of the eye was AT BEST about one minute of arc."

>Given that you can NOT discern any movement of the sun/earth smaller than
>one minute of arc.

>One thing you are missing that everything has resolution...everything, even
>all analog systems.

NO! NO! NO! (sorry for shouting) Measurements of things have resolutions.
The things being measured DO NOT have resolution.

>> This then comes back to my original assertion that all continuous
>> tone photographic mediums have an infinite number of tones
>> whether they are ink jet or silver based.

>And that assertion is flawed.  ALL analog systems have resolution, and
>therefore have a number of discernable tones, and are NOT continuous.

No, analog systems do not have resolution because by definition they are
continuous. The measurements of continuous analog systems do have resolution
because they are not made with perfect instruments and the readings are
expressed in discrete mathematical values.

>> Thank you for your patience if you got this far. <G>

>I have tried, and don't believe I can not explain anything further to you.
>It is not a matter of disagreeing, it is a matter of what you are saying is
>just wrong, and is based in ambiguous definitions and flawed assertions,
and
>you don't want to understand that.

I suppose that if you cannot explain your position within the framework of
standard definitions and must resort to rewriting or refuting them, you may
have no recourse but to simply state, "You are wrong!" which is seldom a
compelling argument.

>I don't know what you do for a living, and what your background is.  I am
>guessing software?

Should have gone into computer science but my degree is in Chemical
Engineering. Five years teaching high school math and science. 17 years in
chemical process and project engineering.

>If you would like to ask another expert in the field, I am very happy to
>suggest someone to you.  Perhaps they can explain it better to you than I
>have been able to.

Unfortunately, I feel I understand the matter reasonably well for my
purposes. I had hoped that providing other sources would help you understand
what I was saying. The dialogue has helped me to clarify and confirm my
thoughts.

Thank you,
Martin