Don said:
I explained all that several times already, including that I'm not
saying everyone must use an external editor. I'm just stating the
facts. It's up to each user to decide what they want or need. If one
is happy with scanner editing tools, more power to them! But if one
says they are equal to an external editor, that's simply inaccurate,
for the many reasons outlined earlier.
I'm not saying that one should not use an external editor because the
driver-integrated tools are equivalent. They are obviously not, with any
dedicated editor having many more tools.
I'm saying that the tools the scanner driver does have (typically
curves, level and a histogram) don't have any specific flaws that make
them worse than what's in Photoshop.
(That's a general statement, of course I suppose you can find a scanner
driver with irremediably bugged curves and levels...)
Any flaws are thus not in the *tools*, but in the fact you're applying
them based on a preview -- which is what we are discussing.
But if you maintain that it's the *scanner driver tools* themselves that
are inadequate, then please explain how they are. The "main reasons
outlined earlier" simply refer to something else.
That's inconsistent. You can't say you disagree and then in the same
breath confirm what I just said (i.e. that what you disagree with is
actually correct).
You said they're massively inaccurate.
I said I they are inaccurate.
Clearly I interpreted you "massively" as meaning "too much, for any
purpose", which is what I do not agree with.
The two key words being "for me". We are not talking about subjective
judgment. As I keep repeating, if it works for you, fine. But that
does not make it objectively true.
Nor objectively false.
Anyway, my "for me" was not really intended as to imply subjective
judgement, although it does probably sound as such.
"For me" = "With my own scanner, they're good enough to set decent (i.e.
not-clipping, but still narrower than 0..255) w/b-point values at first
try in the vast majority of cases. Others may find that they'll often
have to make multiple trial scans (perhaps too many for their likings)
to obtain the same result)".
Because, you see, I maintain that a good result is *always* obtainable:
in the worst case, with a very bad scanner, you'll have to take a number
of "trash scans" before coming out with a good scans.
Now, yeah, *really* having to do *that* could be labeled as "tying
oneself in knots", I admit: you'd come up with a good result, but only
through such a time-consuming process that, I guess, nobody would do it
in practice.
However, the fact that it doesn't get (nearly) to such an extreme with
my scanner makes me infer that it'll be workable with many other scanners.
It may mean you have lower
expectations or any number of other things.
Actually, what I'm trying to say is that I suspect the OP could obtain
*better* results by exploiting the internal 10-bits than by ignoring them.
You may disagree with this statement, but saying that it may mean I have
lower expectations than someone else just does not make any sense.
Nothing wrong with that,
of course. What is wrong, however, is to assume that your personal
preferences translate into objective statements. They don't.
No preferences.
I stand that: if you own a 10-bit external / 8-bit internal scanner,
then by taking a less-than-full-resolution scan ("preview") followed by
one or more full-resolution scans, you can *always* obtain more
information content than by using a "standard procedure" (i.e. just scan).
In particular, here is the complete algorithm, so that no doubts need
remain:
1) Take a scan with with bp=0, wp=255, using any resolution ("preview")
2) Take a full-resolution scan using bp=preview_bp, wp=preview_wp
3) Repeat step 2 if clipping is found
It's quite simple isn't it? Well, I haven't mentioned using a "safety
margin" and this kind of things, but they're not strictly necessary.
*Please* demonstrate what is flowed in the algorithm above. All I can
see from it is that the information content in the final scan will be
higher than by "just scanning" for any given underexposed or overexposed
scan target.
Now, *that* specific algorithm, not having any safety margin, is a bit
impractical to use, since it is bound to force the user to scan more
than once at full resolution.
So, you might find it to slow to use, but now *that's your subjective
opinion and preference*, it's not a "tying myself in knots" on my side.
Anyway, if you keep a margin, it can become much more practical to use,
even though the improvements will diminish.
Meta level diversion:
I think there's a basic misunderstanding of the principle of what I'm
saying. When I'm stating generic facts I'm not advocating any
particular application or use of these facts. I'm just offering
pertinent information. It's then up to each reader to decide what to
do with those facts: ignore them completely, use them fully, use them
only in part, ... etc.
Same for me.
I have! The inexactness of scanner software environment causes vastly
inaccurate settings.
Well, no, you haven't. You have *stated* so, but not, IMHO, satisfyingly
demonstrated so. Please attack the algorithm I've given above,
demonstrating it's flawed.
And "it'll take too many test scans to be practical for people" just
doesn't cut it.
That's just not realistic. The clipping settings are on the order of
0.3% to 0.5.% and you're bracketing with 20%!?
Sorry, I'm not sure what you mean. What are the clipping settings?
Are you saying that I could safely bracket with less than 20%? That
would be even nicer. As I said, I kept on the safe side.
...
Because it starts with a wrong premise and then goes downhill from
there.
What's the wrong premise?
This is a prime example of tying yourself in knots trying to justify
this wrong premise and trying to solve a problem that doesn't exist.
The "problem" here is that the lowest 2 bits from the OP's scanner can't
be obtained directly.
Please explain why this problem doesn't exist.
(The only explanation I could think of is "more than 8 bits per channel
are useless", and other people would give such an explanation, but
you'll have a harder time than them, since I know you wouldn't say that)
If one is so concerned with accuracy, one should simply edit in an
external editor and all of the above "problems" disappear.
And two more bits would appear by magic from nothingness? Nice.
Because you explained it yourself:
No, I didn't.
You "..."'ed the part where I explained why this is no explanation to
your "yes, it is because ..." above.
No, no, no, you misunderstand. What I'm referring to is: You object to
the facts but then immediately add in the following sentence that the
facts are actually true. You do this all the time.
Cite please. At most, I think I might have sometimes been too vague
about exactly *which* parts of your facts I consider true, and which
ones I don't.
On the other hand, I suspect that I have later pinpointed that better
later in almost every case... since you pointed it out in almost every case.
It's inconsistent. So what I'm saying above has nothing to do with
style but with substance which is contradictory.
I see, but I don't think it is.
What do you mean by "defending" a compromise?
For example, your above attempt of trying to come up with a convoluted
clipping margin of 20%.
I'm really missing something here. I mean, you can go with much less
than 20% if you like (and I suspect it would work quite well with many
scanners, very rarely forcing to scanning the same picture more than one).
Actually, I *do* use less than 20% with my scanner.
I just chose 20% because the higher, the "safer" (i.e. there's less risk
of having to scan again, even though there's also proportionally less gain).
None of the above. It's number 3: "There is a perfect solution already
available (edit in an external editor).
Perfect? No. It doesn't let the OP exploit his scanner's internal 10
bits in any way.
You may argue (and you're doing that in fact) that "the cure is worse
than the disease" (which I don't think is true, in any case", but you
can hardly argue that the disease is a "perfect solution".
However, I want to use scanner
software to edit. But it's bad. So I will tie myself in knots trying
to work around its inaccuracies - but not succeeding. I'll end up with
worse results than easily available solution of just using an external
editor. I agree that's the ideal solution,
As I said, I definitely don't, given you have no way to make use of the
lowest two bits in an external editor. Please don't put too many words
in my mouth.
but I will continue to
defend the inferior and inexact attempt at a workaround in scanner
software."
Bah.
Which - as I've shown - is irrelevant because even if he could make
scanner software work at 10 bits internally, due to all the outlined
problems, that would still be worse than using the full complement of
tools on 8-bit data in an external editor.
Yeah, except that I don't think you have shown it, while on the other
hand I do think that I have shown the contrary.
Ah, life, don't we all love it?
But that's simply irrelevant for the reasons I just explained in the
previous paragraph.
You: "The point is he isn't even sure if those two bits are thown away
before scanner software applies any edits."
Me: "I've told you and the OP [..] how to try to find out whether those
too bits are indeed used"
You: "But that's simply irrelevant"
Who's being contradictory now? Is it "the point", or is it "irrelevant"
Oh, wait, I suppose a point might well be irrelevant -- that's not a
contradiction.
So, ok, I recognize that you found out your own point was irrelevant
What's the point of wasting time trying to find
that out when in the end - even if true (!) - the end result would
still be inferior to editing in an external editor?
Except that it wouldn't.
It just doesn't matter. The only thing that would matter is if he
could actually get those 10-bits out. But he can't.
I wonder why they make scanners that have more bits internally than
there are externally, if they serve no useful purpose.
We know the marketing is a son of the devil, but there's a limit to
everything!
[snip]
One final meta level observation:
If what I write sometimes appears "picky" this could be because the
intentions are misunderstood. The intention is *not* to be picky but
to point an (objectively) important aspect. I don't expect this will
be important to everyone (subjectively). However, in that case they
can just ignore it. The problem only occurs when they try to use
subjective assertions to counter objective fact.
I mean, the normal reaction to such a statement offering additional
information (which is my only intention) and made in a
non-confrontational, matter-of-fact fashion (i.e. not argumentative,
but simply and calmly stating facts), a normal reaction to that would
be:
A. Thanks, but no thanks! I don't care for such level of detail
B. Huh! I never thought of that! Thanks!
I'm perfectly OK with either reaction.
I see, but my reaction (*in the case at hand, i.e. the OP's*) is
"C. Sorry, but I think better level of detail can be obtained in another
way".
Clearly I could be mistaken, but it's certainly another valid reaction
in principle.
Note the emphasized "in the case at hand", as in *my own* case (which we
discussed in another thread), my reaction is clearly B (though with
additional notes), as my scanner is perfectly able to scan at 16-bit
external if desired.
by LjL
(e-mail address removed)
)