Don said:
OK, I think we're digressing further and further away from the subject
matter here. You also seem to spend most of the time arguing against
things I never said or implied.
So, it's time to recalibrate (sic) the discussion.
You are right - we are far off topic, and that we should recalibrate the
discussion.
I have never advocated or implied doing away with calibration! And
most *certainly* I have never been a proponent of guesswork! Quite the
contrary!!! My very first message in this forum 3 years ago was asking
for *exact* Kodachrome compensation rather than my own "guesswork", no
matter how good the results of that "guesswork" appeared to be!
Actually, you're argued quite consistently against calibration slides
for reason that don't hold water, which is why I am taking you task on this.
Don, to quote you:
"So, assuming one is after maximum accuracy to that degree, finding a
"one size fits all" calibration slide appears impossible. One may
obtain general hints from a generic calibration slide but not accurate
calibration to that degree of perfection."
So according to you, using calibration slides is counter productive.
You further state:
"I personally, have given up on that approach, and prefer a
self-correcting process whereby type of film becomes irrelevant. This
is not easy, but there are some things which are constant. For
example, the unexposed part of film contains "built-in calibration
data" in that it's supposed to be black (or transparent for
negatives). "
This again, you argue against the use of calibration slides in favor of
your method, which you call "calibration", but can only be loosely
defined as such.
Now then, passive, canned calibration is a compromise.
By that I will take you to mean a calibration measured at a single point
in time rather than on going.
A good
compromise, maybe even a very good compromise, but a compromise
nevertheless.
Let's ask what is being compromised?
1 - the degree to which the calibration slide film batch color balance
matches the film batch I am using. That difference is *very* small.
Smaller if pro film's are used. Check out Wolf Faust's site for futher
data on his tests in this area. According to him, and he's an expert,
the difference is not worth considering.
2 - the extent to which the particular calibration slide matches the
values supplied for that run of slides. To counter this, you can order
individually measured slided for a nominal extra fee. No more problem.
Therefore, dynamic, active, self-correcting calibration will always
produce superior results because it deals with what *is* rather than
with what "should be" - as passive calibration does. Here's a
non-rhetorical question: What will produce more accurate results, a
passive monitor calibration or an active one created with the Spyder?
Obviously a profile created with a spyder is more accurate,
The major problem with your method is that you are saying unexposed film
"should" be black an completely overexposed film "should" be white.
And that the scanner manfucture setting for black and white color
"should" be accurate.
But they are not. Which is why you need to use calibration slides and
colorimeters.
Now, given all that, passive calibration therefore assumes a certain
context and is implicitly limited to that particular (more or less
"narrow") scope. In English, this means such calibration comes with a
bunch of caveats i.e a number of "givens". Passive calibration
"compromises" differences between individual units, it "compromises"
difference between individual media, it "compromises" changes in
scanner over time (long or short term), etc, etc.
So, why isn't everyone using dynamic calibration, then?
Because unless you have a very high end scanner (or monitor) continuous
calibration is impossible.
First of all,
many do without knowing it! My scanner recalibrates itself
automatically, after it's been on a certain amount of time, to
compensate for changes due to temperature built-up and other changes
in the environment. It does *not* depend on a passive, canned
calibration alone. It knows better! What "should be" is not the same
as what *is*.
Yes, but we
1- do no know what the scanner considers "reference" - ie, what is its
black and white point, and how accurately were they determined for
individual scanner (vs for the model line)
2 - do not know what the scanner is correcting - color (which ones?),
exposure (both?) and how accurate or linear the correction is in the
color space.
Yes, its good that the scanner recalibrates, however I know from
experience with 5 different scanners that the self cal cannot be relied
upon to keep you colors acurate over time.
But more importantly, explicit, dynamic calibration is more time
consuming and requires more work. (Do you attach the above mentioned
Spyder to your monitor every 5 minutes, or just depend on occasional
calibration?) Most people couldn't be bothered and prefer the "fire
and forget" of passive calibration.
My slide scanner is always calibrated at the start of every session and
once or twice a day there after depending on the number of scans.
I calibrate the monitor based on the type of work I am doing. If I am
preparing for the web, once a week or less. If am printing for a
portfolio or exbition, once a day or sometimes once every 6-8 hours.
The printer - if for an exhibition (very rare) once each batch of ink
changes - I buy several catrideges from the same batch. Though lately,
they have been suprisingly consistent batch-to-batch.
Otherwise once per session.
Some monitors from Sony, LaCie and others have built in sensors and do
it continously
And that brings us to the last, but perhaps most important aspect,
usage! Depending on intended usage passive calibration may be more
than sufficient and going to time and labor intensive active
calibration will be a case of diminishing returns. Whether that's the
case or not, depends on intended usage in each particular instance.
Obviously, for example, requirements between "scanning for web" and
"scanning for archiving" differ diametrically.
The flip side of usage (implied above but let's make it explicit) is
the level or accuracy! Different applications require different levels
of accuracy. Often times we tend to go to extremes without realizing
the futility of going too far. So, every now and then, it's useful to
step back and ask oneself how much accuracy is "too much" accuracy? In
other words, passive calibration may in many (if not most) cases be
more than sufficient, but that does not mean active calibration will
not produce more accurate results! Double-negative notwithstanding...
;o)
very tru.
So, let's see first where we get with above baseline before atomizing
the discussion into digressing and arcane minutiae. We can do that
later, but I suspect most of them will become moot once we address
this baseline.
All well and good, and there are compromises we make with the frequency
of calibration. That is obvious.
Howver I also won't let you change the discussion to calibration
frequency, because that is *not* where the error in your method lies.
The magnitude of the error introduced by infrequent calibration is far
exceeded by that of not using a proper external reference.
Your method of measuring base fog on the unexposed film and calling it a
black point is flawed. Inherently and irreparably.
The problem is that the film base is an arbitrary reference. There is no
getting around it.
Using an ICC profile created by a calibration slide of a reference color
target is an independent, objective, externally referenced method of
adjusting your scanner's color reprentation to neutral.
That cannot be denied.
If you want ot argue that it is easier to measure the unexposed film and
call it a "black point" than it is to use a calibration slide, that's a
matter of preference.
If you want to say that you get more pleasing results in your work flow
by *not* using a calibration slide, that is up to you.
But you cannot say that you methods produce more accurate, consistent,
reproducible and predictable color output, because it simply does not.
I have to wonder. On the one hand you go to great lengths to argue for
"accuracy" and "objective" quality.
You've argued extensive about how the problems in Vuescan, calibration
and other thing may not matter to most, but are only of interest to
those who are truly demanding and discriminate.
And yet you choose a highly biased, completely inaccurate and wildly
variant color work flow that is little better than an educated guess?
You're trying to pick apples from a banana tree.
Go into any decent pro digital output lab. One that has some good gear -
maybe even a drum scanner and a good Fuji printer. Ask their senior
printer about calibration. It will be an eye opening experience.
If they are not too busy, try this: (or do it at home if you can)
Start with a colorimeter calibrated monitor (very important). Now do 3
slide scans. One with no calibration of any kind. One calibrated with
your unexposed tail. And one calibrated with and ICC profile. Depending
on you software, you may be able to do it with one scan and just apply
the two different profiles.
You will see immediately what I'm talking about.
