ðòïåëôù 


  áòèé÷ 


Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

  óôáôøé 


  ðåòóïîáìøîïå 


  ðòïçòáííù 



ðéûéôå
ðéóøíá












     áòèé÷ :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[filmscanners] Dynamic range... Proposal Part 1


  • To: lexa@lexa.ru
  • Subject: [filmscanners] Dynamic range... Proposal Part 1
  • From: "Julian Robinson" <jrobinso@pcug.org.au>
  • Date: Fri, 23 Aug 2002 10:51:44 +1000
  • Unsubscribe: mailto:listserver@halftone.co.uk

Austin,

I realise I have left a long gap in this discussion,  this does not mean I
have left you to hold the floor.  I have been otherwise occupied.

The problem is what to do now.  I continue to think that the question is
well worth resolving because it affects a great deal of the discussion
here, and as I have said before, I have witnessed an enormous amount of
wasted frustration, discussion and confusion over the years resulting
directly from your mystically complex ideas on this subject.  Simply you
are telling people untruths, and they get confused.

Our discussion is obviously going round in circles, and I have a good
theory as to why this is so.  It is also obviously frustrating a lot of
people (including me) and taking up a fair bit of bandwidth trying to
resolve.  I therefore propose:

1) We choose to focus the discussion.  Let's use the single question /
answer technique.  Will you allow me to ask you a single question, wait for
your answer, then ask another etc until you have convinced me of your point
of view, or vice versa, or even...gasp... we reach a common
understanding?  This will not only focus the discussion, but also mean that
we can be sure to get a response to each point. It will also limit the
number of posts on the topic, and limit the overall time we all spend on
this. It may take weeks at such a slow rate, but only a small amount each
day and I am sure it will be more efficient than what we are doing at the
moment.

2) As you are aware, Todd established an entire group just to discuss this
question.  ( http://groups.yahoo.com/group/max-noise/ ).  This seemed to be
an eminently sensible idea given the distress the interminable
non-converging discussion was causing to some people here.  You were
specifically asked to join max-noise but have chosen not to participate for
some reason.  Would you now agree to take this discussion to that group?

Anybody else have any views? - if the discussion should be here or on the
max-noise group?  If you are sick of the whole thing and don't care about
the outcome then the correct but polite answer is "take it to the other group".

Just the same I think most people would find the following interchange
mildly entertaining or interesting.

Austin I respond to your unnecessarily condescending and arrogant post - in
detail - here...

> >Julian said: The dynamic range is calculated from the Scanner OECF by:
> >
> > DR = Dmax - Dmin                (7.2)
> >
> > DR = Scanner Dynamic Range
> > Dmax = Density where the Signal to noise ratio is 1
> > Dmin = Minimum density where the output signal of the luminance OECF
> > appears to be unclipped
> >
> > -------end quote from Proposed ISO standard-------------------
> >
> > (and OECF is opto-electronic conversion function)
> >
> > You will notice, it is exactly as I have described it, a RANGE.
>
>Austin replied: I do not see ANYWHERE where it says dynamic range is "a"
>range.  It shows
>the RESULT of a calculation WITHIN A RANGE (Dmax), divided by the noise
>(Dmin), but the result is NOT "a" range.

Julian now replies: Austin.  There are several arguments which demonstrate
pretty clearly that dynamic range is a range.  Here are the first two of
them, one of which is discussed  in more detail below, and the rest will
come another time.

1. In your last sentence, you seem to think that Dmax is a range.  It is
not, it is a value.  This single unarguable conceptual error should put the
rest of your assertions to rest, but ... I guess not.

As you know, a range is the "distance" between two values.  Or ... a
dictionary definition: "range  noun [U]  the levels or area within or to
which something is limited "

Thus a range is defined by a top value and a bottom value.  Dmax is one
value, it is not a range.

In some cases it may be reasonable to assume that a value implies a range
with respect to some understood reference value, but that is not how it is
being used in this equation.  Dmax in this equation is the top end (or
bottom end, depending on your point of view) of the range in question - the
range in question being from Dmax to Dmin.  Dmax does NOT itself represent
an implied range from 0 to Dmax, it is just a value, Dmax.

The range that IS mentioned in the quoted equation (taken from the
standard) is the range between Dmin and Dmax - now THAT is a range.  It is
the range between the smallest density and the largest. Or the largest and
the smallest, no matter.  This range is quantified here as a ratio, which
is a very good way of quantifying a range.  In fact it is the ONLY way of
quantifying a range so that the resulting number is independent of gain and
other irrelevant things.  That is why it is used for the definition of
dynamic range -  range = Dmax/Dmin.  Or if we express the densities in log
terms (whether Bels, or decibels or purple turtle doves, it doesn't matter
and I won't insult you by giving you a lesson on Basic Logarithms for
Beginners) it is Dmax - Dmin.  That is a quantification of the range.  It
is the ONLY way you can quantify a range so that it is independent of
irrelevant things.

2.  A small examination of how the language works...

         A turtle dove is a particular type of dove.
         A brick house is a kind of house
         A tall woman is a woman
         A frustrating discourse is a discourse
         A dynamic person is a person
         A superficial argument is an argument
         A dynamic environment is an environment
         Dynamic HTML is a kind of HTML
         A dynamic duo is a duo
         Dynamic games are games
         Dynamic graphics are graphics
         A dynamic force is a force
         A dynamic impedance is an impedance
         A dynamic range is a .....?????

A resolution?     Not likely Austin.  I have struggled to think of one
dynamic 'x' which is not an 'x' and failed, but I admit my intellect may
just be insufficient, and certainly the array of statements above is not
itself a proof in a philosophical logical sense. There could be exceptions
to this implied rule.  However, perhaps you would agree that if dynamic
range is really a resolution and not a range, it would be a very rare
exception to an otherwise pretty good rule?  And if it is such a rare
exception, do you not think that every lecturer when first introducing
dynamic range to their enthusiastic first year engineering students, do you
not think they would say at the very beginning .... "be careful with this
little trickster, it says dynamic range but it is really a resolution!"?

Don't you think that EVERY book on the topic would be at pains to point out
this seemingly illogical internal contradiction?  That EVERY definition of
dynamic range would say "Hey, although dynamic range is called dynamic
range, it is really a resolution"?

Or even more likely, don't you think it probable that technical people in
their enthusiasm to be explicit and clear and make things as easy as
possible would actually have called this thing the "Dynamic Resolution" in
the first place?

But no - the lecturers and the books still call it Dynamic Range, and they
still don't warn their students that it is a misleading term - never, not
once - and they and engineering practitioners still don't set up pressure
groups to have the name changed to Dynamic Resolution and they still don't
mention this bizarre contradiction in their definitions or their standards  ...

Why do you suppose this might be?  Because Dynamic Range is a range, both
in the common English language meaning of the term, and in its technical
meaning and implementation.  The strange and annoying thing about all this
is that if one accepts that dynamic range is a range, then everything that
follows is much simpler, more comprehensible and above all more CONSISTENT
than if you decide, for some reason, that it is a resolution.

QUESTION - I'll ask you again, and I will keep asking until you answer
because it is fundamental to your unique view of what goes on with dynamic
range .... will you please quote me who in this world says in a formal
document (a book, a paper, a standard ,,, anything) that dynamic range is
not a range, it is a resolution.  PLEASE TELL US.  (Note that I am NOT
asking you to interpret a formula that someone has written down, I am
asking you for someone who says that dynamic range is a resolution.  Anyone.)

>It shows (in non-log numbers):
>
>dynamic range = amplitude / noise
>
>and in log math, division is merely subtraction:
>
>Non log example:
>
>1000/10 = 100
>
>log example:
>
>log 1000 = 3 (10**3 = 1000)
>log 10 = 1   (10**1 = 10)
>
>3 - 1 = 2
>
>and 10**2 = 100...fancy that.

Exactly.  Thank you so much Austin for this introduction to logarithms.
Since I have clearly shown in previous posts that my background and
knowledge of logarithms is perfectly adequate for the task of coping with
your complexities, what is your motive here?

> > It is the
> > range between Dmax and Dmin.
>
>No, that's Dmax MINUS Dmin.  That's a MINUS sign there, Julian, it's a
>mathematical equation.

Yes it is a mathematical equation for quantifying the range between Dmax
and Dmin, quite correct.  As you go on to say...

>The result IS the dynamic range.

...and it describes, as a ratio, the signal range possible.  It says, the
signal range possible is from Dmax up to Dmin which we can quantify as the
ratio of Dmax/Dmin.  That is quite true.  As I said above, a ratio is the
ONLY meaningful way you can describe a range in a single figure when you
need to be independent of irrelevant matters like gain.  It is not hard to
understand - 1dB is a small range (about 1.26 to 1), 100dB is a big range
(10000000000 to 1). It's easy.

This is one of those fundamental problems you seem to have.  The inability
to appreciate what is the range we are talking about.

Let's use yet another simple example.  We have an amplifier (assume it is
linear) whose output is 10V max, and the smallest signal we can get out is
one volt of noise when there is 0 input.

We want to measure the dynamic range for this box.  But first, what is the
"common English language RANGE" of output for this device?  Answer - the
output may vary between 1v and 10v.  The RANGE is 1v to 10V.  Now, we want
to express this range as a single figure.  We could say the range is 9v,
because it is, if you want to measure it as a difference.  Again this is
still the common English use of the word range. Unfortunately that way of
measuring is not very useful in a technical environment because the
absolute values of the range, the limits in volts, will depend on where you
measure them and how.  For example, consider the effect of load.  If we use
a different load on this device, the output might fall to half, say 0.5V to
5V, or if we measure the output into an open circuit, the output range
might be double, then being 2V to 20V. In each case, the actual dynamic
range of our device is still the same, but the output voltages are
obviously different.  If we attempt to measure the dynamic range as a
*difference* in each of our three load cases, we would have

a) 5V - 0.5V = 4.5 V range
b) 10V - 1V = 9V range
c) 20V - 2V = 18V range.

We can see that this is not very useful.

So, we decide it is more useful to measure the range as a ratio.  Then the
dynamic range as measured is:

a) 5/0.5 = 10
b) 10/1 = 10
c) 20/2 = 10

... 10 in each case.  Fantastic,  this figure of 10 is a useful way of
measuring the range from the smallest discernable signal to the max signal.
It is 10:1.  That is a measurement of range.  I repeat, yet again, it is
the ONLY way of usefully measuring signal range to avoid being confounded
by arbitrary gain and other factors.  That is why we choose to measure
dynamic range as a ratio.  It is still a range, it is quantified as a
ratio. Can you grasp this concept?

>  It is NOT "the
>range BETWEEN", it is the result OF the subtraction of two log
>values...which, BTW, is exactly the same a DIVISION if the numbers were NOT
>log values.  Think about that.

I don't need to think about the second half of what you say, it is
obvious.   You are saying that the dynamic range is expressed as a ratio. I
agree. See above. You don't need to keep bringing logs up as though they
are a mystery, to you and me they are not, so feel free to dispense with
the censorious lectures on what logs are. If you are only doing it to
convince less technical people that it is very sophisticated stuff then
stop, because it isn't.  The fact that your responses are pitched at such a
simplistic level indicates to me that you have missed the points I was making.

----> continued in part 2 post

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body



 




Copyright © Lexa Software, 1996-2009.