What is better Nikon Super Coolscan 4000 or Coolscan V ED?

  • Thread starter Thread starter Andy Salnikov
  • Start date Start date
Don,

that reply of yours was not actually needed. I guess that many
readers like me, even those who know neither you nor Fred, took
his reply as a confirmation that your message was correct. (:-)

Then, I'm afraid, you misunderstood his response, Hans-Georg. He quite
clearly does not agree and is offensive. Please read it again!

I wrote a very calm and factual message explaining everything in
detail and even offered two improved alternatives (Gaussian Blur and
manual alignment). Everyone else understood it as such (calm and
factual) but Fred was the only one to see it as "anti-Vuescan"!

So, calling my detailed and *neutral* reply "anti-Vuescan" and lashing
out about "whether YOU want to call what Vuescan does as multipass" is
clearly an overreaction and a personal attack of a very angry person.

Therefore, my response you quoted above is both factually correct as
well as appropriate.

Don.
 
Actually, the fact that two successive scans are misaligned
could be used by clever software not only to yield less noise,
but even to yield a higher resolution.

Very time-consuming, but a very interesting way out for those who
cannot afford a scanner with a really high resolution.

That's a very nice lateral thought but as Lorenzo mentioned, the main
problem is that the misalignment is not uniform. It would still be
possible to align them but it requires a lot of very time consuming
processing. And such alignment (due to the way it works i.e.
interpolation) will probably minimize the "useful misalignment" for
increased resolution (interpolation tends to blur the image slightly).

But even if all that worked it would probably not be worth it because
any resolution improvement would be very minor - if any! Namely, in
case of LS-50 and other 4000 dpi scanners, in practical terms, they
already extract the maximum resolution from average film.

A very few shots (using a tripod, high resolution film and perfect
exposure) may have more than 4000 dpi usable data but most shots are
more than adequately covered by 4000 dpi.

BTW, I read here a very long time ago that (since film grain is not
uniform) to literally get even the smallest grain one would need about
10,000 dpi resolution. However, the problem is grain is not two
dimensional but three dimensional ("grain clouds") i.e. film has depth
as well. Therefore, one could only focus on a particular "slice" of
the grain cloud making such extra high resolution of limited use.

Don.
 
You can find Photomatix here:
http://www.hdrsoft.com/

There's also HDRShop which is free:
http://www.debevec.org/HDRShop/
However, it only works with 8-bit input. It will read 16-bit files but
it converts them to 8-bit before processing.
For reference, here are the other methods I've tried for difficult
slides:

1) Use the "long exposure pass" feature with this scanner: this did a
seemingly good job - until I started pixel peeking, that is. What I
found was that it generates some strange artifacts. These are probably
related to mis-alignment between the two scans.

Indeed! That's all I'm saying! If one doesn't pixel-peek it all
appears "great". But if you look carefully everything changes.
2) Multipass approach. This seems to work as it is supposed to. But
the number of scans required in order to see a significant improvement
makes the scanning time just too long (especially with the rather slow
FS4000). I never really convinced myself i was seeing the
mis-alignment induced blurring that people worry about, however.

Need to peek at those pixels again! ;o)

The above "long exposure pass" is also multipass but limited to two
scans only. To see the multipass blurring compare to a single scan at
an appropriate resolution. Or perform multiple scans and merge them in
an editor manually. Then compare this merge to any component scan.

Don.
 
Don wrote:
">that reply of yours was not actually needed. I guess that many
readers like me, even those who know neither you nor Fred, took
his reply as a confirmation that your message was correct. (:-)

Then, I'm afraid, you misunderstood his response, Hans-Georg. He quite
clearly does not agree and is offensive. Please read it again! "

Actually, Don, I think you misunderstood what Hans-Georg was saying. I
believe he was saying that when Fred responed with an ad-hominem
attack, and not a substantive reply, that it was clear that it was
because you were right. That is why you didn't need to reply- any
third party would clearly see this dynamic. Subtext is key.

You're right about the limitations of Vuescan multipass scanning and
it's too bad this isn't documented in the official user-manual and only
strewn about newsgroups. Where would I be without this group to
explain how VS really works?
 
Actually, Don, I think you misunderstood what Hans-Georg was saying. I
believe he was saying that when Fred responed with an ad-hominem
attack, and not a substantive reply, that it was clear that it was
because you were right. That is why you didn't need to reply- any
third party would clearly see this dynamic. Subtext is key.

I understood Hans-Georg's subtext to mean that even though I was
right, I went too far with my reply to Fred. Specifically, it was
H-G's phrase "that reply of yours was not actually needed". You know,
as in "that was way over the top" or "easy now, no need for that".

But you're absolutely right, that phrase could be interpreted in two
different ways: 1. My reply is redundant, or 2. My reply goes too far.
You're right about the limitations of Vuescan multipass scanning and
it's too bad this isn't documented in the official user-manual and only
strewn about newsgroups.

I suppose to document it would mean admitting its, shall we say,
"limited usefulness". Unfortunately (to put it mildly) the VS author
is not exactly known for being up front about the program's
shortcomings. How's that for understatement? ;o)
Where would I be without this group to explain how VS really works?

Not only VS but everything else regarding scanning! This group is a
veritable treasure chest!

Don.
 
"Unfortunately (to put it mildly) the VS author is not exactly known
for being up front about the program's shortcomings. How's that for
understatement?"
Very subtle and yet very clear ; )

"Not only VS but everything else regarding scanning! This group is a
veritable treasure chest! "

And on that we can completely agree! I've learned a lot here.
 
Roger said:
You're right about the limitations of Vuescan multipass scanning and
it's too bad this isn't documented in the official user-manual and only
strewn about newsgroups.

Well, I quote form the VS user's guide, section on
"Maximizing Image Quality":

<quote>
The second technique is multi-pass multi-scanning, which most scanners
are capable of (however, some can't accurately reposition each scan
pass,
so this sometimes doesn't work well).
</quote>

Still think it is not documented?
(BTW, it works a treat on the Epson 4990.)

Where would I be without this group to
explain how VS really works?

To a certain extent, yes. But sometimes I wish folks here
got on with reading the users guide in its latest incarnations.
VS goes through new releases almost every week and the manual
is upgraded often as well. Perhaps reading the latest before
making a sweeping statement might be more appropriate?

After all I can say a lot of bad things about Windows 3.1 as
well as Linux kernel 2.2. But that hardly matters nowadays,
does it?
 
Roger said:
You're right about the limitations of Vuescan multipass scanning and
it's too bad this isn't documented in the official user-manual and only
strewn about newsgroups. Where would I be without this group to
explain how VS really works?
Are you sure that's right? It's been a while since I used Vuescan or
had any reason to read the documentation, but I distinctly remember
something in the documentation mentioning the problem of multipass and
actually naming a scanner that functioned particularly poorly in this
regard.
 
VS goes through new releases almost every week and the manual
is upgraded often as well. Perhaps reading the latest before
making a sweeping statement might be more appropriate?

One of the problems with VS is that it trips over itself frantically
trying (but usually not succeeding) to "correct the corrections" in a
recursive feedback loop. Judging by many users' complaints that often
only makes matters worse requiring even more rash "corrections".

Staying on top of all that is not an easy task for the user when even
the author himself can't keep up - as he repeatedly demonstrates.

Therefore, blaming the user for such a sorry state of affairs is not
exactly fair. In such a case the blame lies squarely with the author.
After all I can say a lot of bad things about Windows 3.1 as
well as Linux kernel 2.2. But that hardly matters nowadays,
does it?

Not really the same thing because complaints are about current VS
versions and the dynamic is totally different.

You'll be hard pressed to find a Linux user looking for old kernel
versions (which are freely available, BTW) because the new version is
worse. That happens with VS all the time. To add insult to injury the
VS author quickly deletes old versions in an attempt to try and hide
embarrassing bugs.

When a Linux fix is released it's only after comprehensive testing and
quality control. Because of that, as a rule, the bug stays fixed.

When a VS "fix" is rushed out, at best it only hides the bug(s) for a
release or two, at worst it also breaks half a dozen other things
requiring even more panicked "fixes".

Therefore, the flood of VS "releases" is not because the author is
responsive (as is the case with Linux) but because he's desperately
trying to put out fires only to end up starting more. That's why VS is
a house of cards while Linux is a rock, by comparison.

Don.
 
Here's the text from the VS user-guide:

"There are several ways of getting multiple image samples. The first of
these is single-pass multi-scanning. Some scanners are capable of
reading each pixel position multiple times before advancing the scan
head to a new position. The film scanners that can do single-pass
multi-scanning are the Minolta QuickScan 35, Scan Dual, Scan Dual III,
Scan Multi, Scan Multi Pro, Scan Speed, Scan Elite, Scan Elite II and
Nikon LS-2000/LS-4000/LS-8000. The second technique is multi-pass
multi-scanning, which most scanners are capable of (however, some can't
accurately reposition each scan pass, so this sometimes doesn't work
well)."

I really don't think this is adequate documentation as it doesn't tell
someone new to scanning what to look for to know if it isn't working
well. I eventually figured it out by doing tests- change one variable
at a time and compare 4000dpi scans in Photoshop, but this is very
time-consuming. Ditto for not explaining the limitaions of the
"long-exposure pass" which has the alignment and other problems.

How about the limitations to its IT8 profiling, specifically issues
Vuescan has with non-Vuescan created profiles? (try assigning the same
non-VS created profile to a raw scan in VS and Photoshop and convert to
AdobeRGB- I got quite different results)

I'm thinking about putting together a real world guide to image quality
with Vuescan compiling some of what I've learned over the last year and
a half.
 
Lorenzo J. said:
Kennedy said:
Hans-Georg said:

Yes, it is possible - it is a technique I have used myself in imaging
systems, albeit with a deliberate and systematic misalignment between
successive frames.

[snip]

I don't know of any software that permits the automatic alignment of
frames within fractional pixel pitches for interleaving or the decision
on which frames are aligned well enough with any others in the set to be
integrated with each other.

ALE can do all that, as well as perform more geometrical transformations
than just translation in order to get images aligned.
I don't believe it can do the second requirement - which frames should
be integrated and which should be interleaved.
 
Lorenzo J. said:
I'm not sure the two can be obtained *together*, though. I think there's a
bit of Heisenberg involved ;-)

That depends on the ultimate output format. Simply realigning images on
a higher resolution matrix also provides many of the benefits of
multiscanning when viewed at full resolution from the appropriate
distance. It is quite amazing what the eye can achieve.
Yes, two perfectly aligned scans won't do it. But it's dubious to me whether
you can actually achieve higher resolution even with misaligned scans

That isn't in any doubt, and I have built enough systems that have
proved it in side by side comparison to be convinced. The question is
whether the resolution is limited by the native sampling resolution (in
which case misalignment can offer a resolution gain if treated
appropriately) or by the system MTF (in which case it can't). Most
flatbeds fall into the latter category because they are already
utilising a form of misaligned resolution enhancement through the
HyperCCD approach. But many dedicated film desktop scanners fall into
the former category, which may make it worth the effort.
 
Here's the text from the VS user-guide:

"There are several ways of getting multiple image samples. The first of
these is single-pass multi-scanning. Some scanners are capable of
reading each pixel position multiple times before advancing the scan
head to a new position. The film scanners that can do single-pass
multi-scanning are the Minolta QuickScan 35, Scan Dual, Scan Dual III,
Scan Multi, Scan Multi Pro, Scan Speed, Scan Elite, Scan Elite II and
Nikon LS-2000/LS-4000/LS-8000. The second technique is multi-pass
multi-scanning, which most scanners are capable of (however, some can't
accurately reposition each scan pass, so this sometimes doesn't work
well)."

I really don't think this is adequate documentation as it doesn't tell
someone new to scanning what to look for to know if it isn't working
well.

Come off it Roger, it states quit clearly that some scanners aren't up
to the job. It's a user manual, not a treatise on the relative merits
of image processing techniques. It isn't a requirement of the software
vendor to explain something as trivial as the consequence of inaccurate
misalignment - there is some assumption of user competence.
 
I really don't think this is adequate documentation as it doesn't tell
someone new to scanning what to look for to know if it isn't working
well. ....
How about the limitations to its IT8 profiling, specifically issues
Vuescan has with non-Vuescan created profiles?

While it is to be expected that manufacturers don't voluntarily
advertise problems with their products (just obfuscate enough to stay
within the law) Vuescan is, indeed, in a category of its own, in this
respect. This goes far beyond the usual "caveat emptor".

The infamous Vuescan stripes with Minolta 5400 come to mind as a prime
example. There was absolutely no mention of this major Vuescan bug in
the documentation and the author persisted that Vuescan "worked" with
the 5400 for two years running in spite of numerous user complaints
screaming for their money back.

Continuing to maintain "Vuescan works with 5400" in spite of ample
proof to the contrary and failing to come clean about the bug in the
documentation was not only immoral but plain illegal. As was the
author's refusal to refund the money.

So, all the issues you raise are definitely "in character" and do not
come as a surprise.
I'm thinking about putting together a real world guide to image quality
with Vuescan compiling some of what I've learned over the last year and
a half.

That's commendable and it will surely help others but isn't that the
responsibility of the manufacturer? I mean, user-to-user support is
fine and reputable companies even offer web pages for users to
exchange tips, but when you pay for something there are minimal
requirements the manufacturer has to meet. The issues you raise (as
well as many others) squarely fall within this remit.

BTW, if you do write such a guide and dare to expose any Vuescan
"problems" expect vitriolic hate mail from the author (and he doesn't
shy from obscenities either). Judging by history, he'll probably also
tell you that you've been "blacklisted". It's not quite clear what
that means but it's some sort of delusional "excommunication"!?

Don.
 
Well perhaps so- it's not a requirement that the software vendor
provide this information, but this program is documented less well than
others I've used. If there were books on VS the way there were books
on Photoshop that would take the burden off of Ed, but with that said,
Adobe's documentation for how to use Photoshop is a million times
better than the cryptic guide that came with Vuescan.

That this program support so many features but not all work equally
well with every scanner made me wish, when I was a new Vuescan user,
that under the *supported scanners* description there were honest
caveats that X feature is not recommended with this scanner, instead of
leaving it to the user to figure out. You figure Ed knows all of this
from testing each scanner- why not document it for the benefit of the
users?
 
Kennedy said:
Lorenzo J. said:
Kennedy said:
Kennedy,

[snip]

know of any software that permits the automatic alignment of
frames within fractional pixel pitches for interleaving or the decision
on which frames are aligned well enough with any others in the set to be
integrated with each other.

ALE can do all that, as well as perform more geometrical transformations
than just translation in order to get images aligned.
I don't believe it can do the second requirement - which frames should
be integrated and which should be interleaved.

I think it can. It outputs a "match percentage" number for every image, and
you can set a threshold on that to decide whether to keep or discard an
image.

I'm going by memory though, I'd have to check the manual.


by LjL
(e-mail address removed)
 
Kennedy said:
[snip]
Yes, two perfectly aligned scans won't do it. But it's dubious to me
whether you can actually achieve higher resolution even with misaligned
scans

That isn't in any doubt, and I have built enough systems that have
proved it in side by side comparison to be convinced.

Kennedy, you snipped the part where I said *in what context* I don't think
resolution can be gained...

You say you have *built* systems where it works. Built. What kind of systems
were they? As I said in the part you snipped, normal consumer-grade scanner
produce significant amounts of misalignment *that varies from scanline to
scanline*. That won't happen in, say, a digital scanner, or a
high-precision scanner system you may have built.

How is any software going to correct for misalignment that, for all intents
and purpuses, can vary at every pixel? If it does *not* vary, then it's a
whole different thing. But in scanners people normally have in this
newsgroup, it does vary.
The question is
whether the resolution is limited by the native sampling resolution (in
which case misalignment can offer a resolution gain if treated
appropriately) or by the system MTF (in which case it can't). Most
flatbeds fall into the latter category because they are already
utilising a form of misaligned resolution enhancement through the
HyperCCD approach. But many dedicated film desktop scanners fall into
the former category, which may make it worth the effort.

I think that's a big "may". I mean, it's certainly something that would be
interesting to try, I just think the probabilities of success are very low,
for the reasons I mentioned above. But then, perhaps some "dedicated film
desktop scanners" have really really good mechanics, who I am to know. (Or
rather, what money do I have to know ;-).


by LjL
(e-mail address removed)
 
Well perhaps so- it's not a requirement that the software vendor
provide this information, but this program is documented less well than
others I've used. If there were books on VS the way there were books
on Photoshop that would take the burden off of Ed, but with that said,
Adobe's documentation for how to use Photoshop is a million times
better than the cryptic guide that came with Vuescan.

That this program support so many features but not all work equally
well with every scanner made me wish, when I was a new Vuescan user,
that under the *supported scanners* description there were honest
caveats that X feature is not recommended with this scanner, instead of
leaving it to the user to figure out. You figure Ed knows all of this
from testing each scanner- why not document it for the benefit of the
users?
As a general rule, making the product itself is only a small part of the
activity and documenting it can be far more expensive. The more
detailed that documentation is, the more expensive it becomes.

I am quite sure that if a majority of Vuescan users asked Ed to document
the software as thoroughly as Photoshop he would oblige - and price the
product accordingly!
 
Lorenzo J. Lucchini said:
Kennedy said:
[snip]
Yes, two perfectly aligned scans won't do it. But it's dubious to me
whether you can actually achieve higher resolution even with misaligned
scans

That isn't in any doubt, and I have built enough systems that have
proved it in side by side comparison to be convinced.

Kennedy, you snipped the part where I said *in what context* I don't think
resolution can be gained...
Having looked at your previous posting again Lorenzo, there doesn't seem
to be any additional context that I snipped. The text I quoted was your
in-line reply to Hans-Georg's statement - and I specifically kept the
part about correct alignment no providing any gain to keep your comment
in context. You did make additional comments over and above that, but
see below.
You say you have *built* systems where it works. Built. What kind of systems
were they?

Mainly military scanned imaging sensor systems, but one or two
industrial imaging sensor systems I designed used a systematic
misalignment between frames to achieve increased resolution, which was
measured in test rooms.
As I said in the part you snipped, normal consumer-grade scanner
produce significant amounts of misalignment *that varies from scanline to
scanline*.

Yes, I snipped because it was "in addition to" your original comment as
covered by your "come to think of it" statement, not a contextual
qualifier.
That won't happen in, say, a digital scanner, or a
high-precision scanner system you may have built.
Don't bet on it - military kit is reliable, but the precision is often
no better than commercial, particularly when the kit is operating at the
extreme of its mechanical design envelope, such as on a fast jet pulling
maximum g or a tank manoeuvring over rough terrain.
How is any software going to correct for misalignment that, for all intents
and purpuses, can vary at every pixel? If it does *not* vary, then it's a
whole different thing. But in scanners people normally have in this
newsgroup, it does vary.
The question is whether it varies significantly enough to prevent
consistent alignment to half a pixel accuracy or so. That isn't so
demanding as your overarching statement, and many scanners are indeed
capable of this. Even for those where frame to frame alignment might be
out by more than a pixel, making dumb multipass multiscanning (as in
Vuescan) worthless, the variation across the field on each frame can
easily be less than half a pixel.
I think that's a big "may". I mean, it's certainly something that would be
interesting to try, I just think the probabilities of success are very low,
for the reasons I mentioned above. But then, perhaps some "dedicated film
desktop scanners" have really really good mechanics, who I am to know. (Or
rather, what money do I have to know ;-).
Let's just say that the alignment of consecutive scans on my Nikon
LS-4000 is certainly better than half a 4000ppi pixel across the frame,
although the frame to frame alignment itself certainly is a lot worse.
There ay well be long term variations, say if one scan was performed
several hours after the other, but consecutive scanning is pretty good.
 
I understood Hans-Georg's subtext to mean that even though I was
right, I went too far with my reply to Fred. Specifically, it was
H-G's phrase "that reply of yours was not actually needed". You know,
as in "that was way over the top" or "easy now, no need for that".

But you're absolutely right, that phrase could be interpreted in two
different ways: 1. My reply is redundant, or 2. My reply goes too far.

Don,

sorry for being ambiguous, but I did indeed mean that your
message in question was correct, and Fred's unwarranted attack
on it only confirmed that he had nothing substantive to reply.

I usually take such attacks as a confirmation of the attacked
message.

Hans-Georg
 
Back
Top