Minolta 5400 or Coolscan 5000

[Kennedy McEwen]

Very, very interesting talk going on here - thank you very much
everybody for sharing your knowledge.

There is one thing that isn't yet clear to me: What actually is a RAW
scan? The definition seems to vary between users. And more
importantly, how do actually get a raw scan from a coolscan (V or
4000)? As most of you will know, the manuals aren't of any help.

Can someone here please enlighten me with the actual settings (incl.
preferences) needed in NikonScan 4 to get the most raw scans.

I don't know how Don does it, but:

1) in Preferences\Color Management, uncheck "Use Nikon Color Management
System" - you will get a warning that this will not come into action
until you reboot Nikonscan.
2) in Preferences\Gamma set Gamma to 1.00
3) in Preferences\Advance Color, set the white point target to 255, the
Grey point target to 128 and the Black point target to 0 for Master, red
green & blue respectively.

Close and restart Nikonscan.
In the Tools Palette:
4) Uncheck (ie a cross) in the Curves, Color Balance and Unsharp Mask
boxes.
5) Under DICE, uncheck "Enable Digital ICE" and "Enable Post
Processing".
6) Under Analog Gain, press Reset.
7) Disable "Scan Image Enhancer".
8) Under Scanner Extras, select Pixel Data Size as 14 bit.

Finally, press "Preview" and check that the image on the Natural and
Processed pages are identical - if not, go back and check that you
haven't missed any of the steps above. Then Scan.

If you wish to do this as a standard workflow, then press Settings\Set
User Settings.

Of course, if you really want to avoid any automatic scanner function at
all, then you will turn off Auto-focus and Autoexposure for both
positive and Negative film under Preferences\Single Scan.
Preferences\Batch Scan and Preferences\Preview Settings. The latter is
not compulsory since it only affects the preview but it is handy so that
you see what you are about to get.

Personally, I can't see the point, but others differ.

 

[Don]

No I did not delete *the* relevant part of your statement. I deleted *a
part* which was irrelevant to the point I disagreed with.

Which translates into you deleted *the* relevant part because your
whole disagreement is based on the false part you did quote - if you
know what I mean...

Anyway, have it your way... I'm going back to scanning subjects. We
should both get to scanning subjects.

fx: Everyone, altogether now: Thank you, Don! ;o)

Don.

 

[Don]

Well, you don't need to look far for evidence to refute that, your first
input on the thread was:

Toby wrote:
"> I prefer to scan raw and edit in Photoshop.

You lose a lot of quality this way."

Don wrote:
"Forgive me, but that is just plain nonsense. Exactly the opposite is
true.

I said *in context*.

At no point in this, or in most of your subsequent posts have you even
suggested that "any workflow he chooses is his own business. "

How about the *only* part from the same message - and the only
relevant part - you conveniently omitted:

Both VueScan and SilverFast are "point-and-shoot" programs similar to
disposable cameras. OK for casual tourist but not for people who care
for quality or have an inkling of what they are doing. Nothing wrong
with that if that's what they are after, but...

In particular: "Nothing wrong with that if that's what they are
after".

And again:

As I said many times, there's absolutely nothing wrong with that. If
you're happy with what you're getting - even if SF or VueScan or
whatever "massaged" scanner data to death - all that doesn't matter.
The important thing is that you are getting results which satisfy your
requirements.

In particular: "As I said many times, there's absolutely nothing wrong
with that" and "The important thing is that you are getting results
which satisfy your requirements".

So, whenever it was *pertinent* I surely did point it out.
Unambiguously!

Don.

 

[Don]

I don't know how Don does it, but:

1) in Preferences\Color Management, uncheck "Use Nikon Color Management
System" - you will get a warning that this will not come into action
until you reboot Nikonscan.
2) in Preferences\Gamma set Gamma to 1.00
3) in Preferences\Advance Color, set the white point target to 255, the
Grey point target to 128 and the Black point target to 0 for Master, red
green & blue respectively.

Don also sets Black and White clipping to 0%! Very important!

Additionally, Don would also make sure that the default file format is
TIFF and not JPEG! ;o)

Close and restart Nikonscan.

Before restarting, Don turns off the scanner too! As Don posted a
couple of months ago - at least in relation to LS-30 and NikonScan 3.1
- there are still some settings left in the scanner RAM which cause
unpredictable and spurious results.

So, to be on the safe side, Don closes NikonScan, turns off the
scanner, pauses a few seconds for capacitors to drain, then turns the
scanner back on, and only then restarts NikonScan.

Also, Don has observed that VueScan has a similar (but much worse)
effect on the scanner, so Don would - first of all, avoid using
VueScan - but should one start VueScan by mistake, it is strongly
advised to turn the scanner off before starting proper, non-beta
software such as NikonScan... ;o)

In the Tools Palette:
4) Uncheck (ie a cross) in the Curves, Color Balance and Unsharp Mask
boxes.
5) Under DICE, uncheck "Enable Digital ICE" and "Enable Post
Processing".

Don would also uncheck Digital DEE. Since Kennedy doesn't have LS-50
it's understandable he would not know about this and Don doesn't blame
him for it because Don is always fair, objective and balanced - modest
too! - not to mention Don has a good sense of humor! ;o)

6) Under Analog Gain, press Reset.
7) Disable "Scan Image Enhancer".
8) Under Scanner Extras, select Pixel Data Size as 14 bit.

Don would also remove any cropping because NikonScan sometimes gets
confused and tries to recalculate cropping - even if Scale is
displayed as 100%!!!

Sometimes - internally - this is actually 99.999 or whatever which
causes interpolation. Don has wasted weeks on this (see Weird
Histogram thread of a few months ago), has hex dumps of settings files
to prove it as well as lots of pulled hair! ;o)

Finally, press "Preview" and check that the image on the Natural and
Processed pages are identical - if not, go back and check that you
haven't missed any of the steps above. Then Scan.

If you wish to do this as a standard workflow, then press Settings\Set
User Settings.

Of course, if you really want to avoid any automatic scanner function at
all, then you will turn off Auto-focus and Autoexposure for both
positive and Negative film under Preferences\Single Scan.
Preferences\Batch Scan and Preferences\Preview Settings. The latter is
not compulsory since it only affects the preview but it is handy so that
you see what you are about to get.

Don thinks it helps to turn Auto Exposure off - at the very least
before the first, preliminary scan - because Auto Exposure will modify
Analog Gain internally but will (falsely) display it as flat because
"Reset" means Analog Gain is considered flat *after* Auto Exposure
Analog Gain has been applied. (Afterwards, if the histogram of the
full scan is not extreme in any way, Auto Exposure may be turned back
on as long as Clipping is off.)

This can actually be used as a neat trick to push Analog Gain beyond
the irrational NikonScan limit of +/-4 (Master + Individual RGB). By
using Auto Exposure, Analog Gain may first be boosted a certain amount
and yet the display will still show 0! Therefore, the cumulative Auto
Exposure Analog Gain plus any subsequent manual Analog Gain may be
used to push and exceed the nominal limit of +/-4.

Personally, I can't see the point, but others differ.

Don thinks there is a lot of point, because relying on the Preview
"keyhole" leads to inaccurate scans. A Preview scan is a (very!) low
resolution scan purported to be "representative". As a result of this
low res scan the actual scan at full resolution will vary from the
Preview scan. How important this variation is depends on the user. For
the casual user, this will be of no consequence. For the
discriminating user (e.g. someone who is interested in scanning raw)
this is a major difference.

Furthermore, Don thinks that using a limited subset of image editing
tools provided in NikonScan is not sufficient for any serious image
editing especially when trying to get the most out of the scanner. An
example of this can be found on page 42 in the NikonScan manual:

The histogram uses eight-bit precision...
....Sixteen bit precision is however used when the image is scanned.

Also, Preview image depends on monitor profile, and should this be
incorrect or off (as in "incorrect") so will the preview (page 80).
Preview gamma is solely for display purposes and doesn't apply to
image data (page 82). Etc. Etc. Etc.

The inevitable conclusion is that relying of Preview is to be avoided
at all costs when trying to get the most out of the scanner! Of
course, Preview is perfectly alright for a casual user, but the
context of the question relates to raw scans and by implication to
trying to get the purest (most) data out of the scanner. Don belongs
to the 2nd group.


May Don now stop referring to himself in the third person? ;o)

Don.

 

[Don]

On the contrary Don, you have focussed on one specific sentence of
Toby's comment, ignoring the remainder of the paragraph in which he
explained why he believed that to be true. Strangely, you seen to have
a problem when others focus on one particular sentence of yours, which
was much more absolute in its coverage.

Are you referring to this:

You lose a lot of quality this way. You should get s/w that allows you to
get optimized scans into Photoshop in the first place. Doing the corrections
in PS later from less than ideal scans gets you gappy histograms and
sharpening halos, among other things.

Because if you are, that doesn't change a thing, other than to confirm
the vastly mistaken notion that scanning raw "loses quality". The
charitably named "optimized scans" are nothing else but scans
*corrupted* by scanner software "adjustments".

Stating that editing raw images later in Photoshop "gets you gappy
histograms and sharpening halos" shows an utter lack of understanding.

I'm surprised that you of all people, Kennedy, would be relying on
such statements.

And I still don't see how does Analog Gain fit in here? No matter how
you cut it, it still boils down to comparison of "raw" vs. "cooked"
and the above paragraph clearly shows a very elementary lack of
understanding of what that really means.

Don.

 

[Kennedy McEwen]

Are you referring to this:


Because if you are, that doesn't change a thing, other than to confirm
the vastly mistaken notion that scanning raw "loses quality". The
charitably named "optimized scans" are nothing else but scans
*corrupted* by scanner software "adjustments".

This is precisely where I do disagree. Your use of the term "corrupted"
indicates that some damage has been caused to the scanned image. This
is not always, or even generally, the case. In almost 100% of scanned
examples the "raw" image is not suitable as a final result and requires
"adjustements". Implementing that "adjustment" in Photoshop does not
guarantee any superiority of result over implementing the same
adjustment in the scanner software. On the contrary, Photoshop's limit
of 15-bit precision in its internal data handling means that a properly
written "adjustment" procedures for a 16-bit scanner will actually be
superior to the workflow you suggest. Specifically then, it can often
be the case that it is better to implement some processes, eg. curves &
gamma correction, in the scanner or third party software prior to
transfer to Photoshop.

There may be many other processes where Photoshop doesn't actually
implement exactly what it is supposed to as well as the scanner software
does, however verifying which is best would require some analysis of
either the codes or the precise results. You have no evidence
whatsoever that the scanner adjustments "corrupt" the scan any more than
Photoshop adjustments do. And whilst it may well be that you still have
your raw original to go back to and process with a different workflow,
it is the end result that matters.

Stating that editing raw images later in Photoshop "gets you gappy
histograms and sharpening halos" shows an utter lack of understanding.

By you - I believe that Toby has qualified his statement specifically to
8-bit scans. However it is equally relevant to true 16-bit scans given
the known limitations of PS.

I'm surprised that you of all people, Kennedy, would be relying on
such statements.

I am surprised that you are surprised. You have chosen to object to a
refinement of your inaccurate statement. I have never stated that you
were completely wrong on the matter, just not 100% correct in your
absolute terminology. "Exactly the opposite" workflow is not always
superior.

And I still don't see how does Analog Gain fit in here? No matter how
you cut it, it still boils down to comparison of "raw" vs. "cooked"
and the above paragraph clearly shows a very elementary lack of
understanding of what that really means.

As I have made clear in several post throughout this thread, analogue
gain is just one aspect that happened to be the first process that
struck me as not being consistent with your statement that "exactly the
opposite" of Toby's was true but, as I mentioned very early on in the
thread, it is not a unique process that would be inconsistent with your
statement.

 

[Kennedy McEwen]

I said *in context*.

How much more "in context" can it be than your precise and immediate
response?!!?

 

[Kennedy McEwen]

Which translates into you deleted *the* relevant part because your
whole disagreement is based on the false part you did quote

No, it translates into exactly what I said - I deleted *a part* which
was irrelevant to the point disagreed with. The rest of your statement
did not qualify, restrict or change the meaning of the part I
specifically disgreed with, in fact it referred to a completely
different matter. As such it was irrelevant to the point in question.

 

[Kennedy McEwen]

Don thinks there is a lot of point, because relying on the Preview
"keyhole" leads to inaccurate scans. A Preview scan is a (very!) low
resolution scan purported to be "representative".

As I have mentioned several times throughout this thread, the preview
need not be a "very!" low resolution scan - indeed it can be a very
significant proportion of the full resolution. Certainly the default
preview resolution is likely to miss specular highlights in an image,
however you can scale the preview to whatever size, up to almost the
full screen, you want. With such a preview size, it does not make any
significant difference to the final image at all that the "missing
pixels" that Don is concerned about might actually contain specular
highlights that would otherwise be saturated - and I challenge him to
provide a real example of where it actually does!

Furthermore, Don thinks that using a limited subset of image editing
tools provided in NikonScan is not sufficient for any serious image
editing especially when trying to get the most out of the scanner.

Nobody has yet, other than yourself, suggested that it might. However
it makes no detrimental difference to the final result that these tools
are used to pre-process the image rather than implementing the same
processes with similar tools in Photoshop. Again, I challenge Don to
provide a single practical example of Photoshop implementing the same
function to the same level with a superior result to using the Nikonscan
tools.

An
example of this can be found on page 42 in the NikonScan manual:

The histogram uses eight-bit precision...
...Sixteen bit precision is however used when the image is scanned.

By contrast, the Photoshop histogram also uses 8-bit precision - but
only 15-bits on the image.

Also, Preview image depends on monitor profile, and should this be
incorrect or off (as in "incorrect") so will the preview (page 80).
Preview gamma is solely for display purposes and doesn't apply to
image data (page 82). Etc. Etc. Etc.

The inevitable conclusion is that relying of Preview is to be avoided
at all costs when trying to get the most out of the scanner!

No, the inevitable conclusion is to make sure that your profiles are set
the same for your scan & preview software as for your editing software.
It isn't difficult - just check the filenames of the profiles in
question. Once checked you need never worry about them again unless you
have to rebuild all or part of the system.

 

[Hecate]

No, it translates into exactly what I said - I deleted *a part* which
was irrelevant to the point disagreed with. The rest of your statement
did not qualify, restrict or change the meaning of the part I
specifically disgreed with, in fact it referred to a completely
different matter. As such it was irrelevant to the point in question.

Kennedy, just used your post because it was the last in the sequence
when I downloaded...

OK, guys, whilst this is very interesting, can I just jump in with
wishy-washy liberal sit-in-the fence view and say that:

I prefer to scan as RAW as possible *with the minimum amount of
adjustments in the scanner software concomitant with getting the best
scan".

Or, to put it another way - I use the software but only to maximise
scan quality.

Or is that not black and white enough? ;-)

 

[Don]

I prefer to scan as RAW as possible *with the minimum amount of
adjustments in the scanner software concomitant with getting the best
scan".

Or, to put it another way - I use the software but only to maximise
scan quality.

As Inspector Cluseau would put it:

Zet's vat I sed! ;o)

Or is that not black and white enough? ;-)

On my uncalibrated LCD monitor it has a distinct dark, blue cast...

Oh no, wait... That's the Kodachromes... ;o)

Don.

 

[Don]

How much more "in context" can it be than your precise and immediate
response?!!?

By *not* omitting the *only* pertinent paragraph! You ignored it in
your response yet again, so:

Both VueScan and SilverFast are "point-and-shoot" programs similar to
disposable cameras. OK for casual tourist but not for people who care
for quality or have an inkling of what they are doing. Nothing wrong
with that if that's what they are after, but...

In particular: "Nothing wrong with that if that's what they are
after".

You said I never include such qualifications, and yet here is a clear
and unambiguous example which directly contradicts your unsupported
assertion.

Don.

 

[Don]

As I have mentioned several times throughout this thread, the preview
need not be a "very!" low resolution scan - indeed it can be a very
significant proportion of the full resolution.

It just doesn't work that way:

1. If you preview the full image then you only have very limited
resolution.
2. If you preview using full resolution then you only have a fraction
of the image.

You just can't have both *full* resolution and *full* image in the
Preview window. You have to chose one or the other. At least until we
get monitors with 5400 x 3600 resolution, by which time the scanner
resolution will, not doubt, go up as well...

Therefore, you only get a fraction of the image data (either very low
resolution or, a tiny portion of the image) which is why I call the
Preview window a "keyhole".

Sure, you can do multiple previews at full resolution on different
parts of the image, but that would be a cure worse than the disease
because you wouldn't be saving any time that way and just get into
another quagmire.

Nobody has yet, other than yourself, suggested that it might. However
it makes no detrimental difference to the final result that these tools
are used to pre-process the image rather than implementing the same
processes with similar tools in Photoshop. Again, I challenge Don to
provide a single practical example of Photoshop implementing the same
function to the same level with a superior result to using the Nikonscan
tools.

Again, and for the umpteenth time, I never said Photoshop *algorithms*
were superior. Point out where I said that!

All I said was that the scanner software *environment* (the tiny
preview window, the lack of numerical displays, the lack of additional
tools such as Threshold, etc.) is what makes editing in scanner
software unsuitable for anything but casual use.

If that's what you're after, then it's perfectly suitable. But if you
are after maximum quality, it doesn't even come close.

By contrast, the Photoshop histogram also uses 8-bit precision - but
only 15-bits on the image.

Which is why there are plug-ins like Wide Histogram which will even
export histogram data for further analysis in Excel, for example. This
is impossible to retrofit into scanner software which is limited to a
very narrow subset of underpowered tools.

Again, perfectly suitable for a casual user but a nightmare for high
quality data acquisition.

Don.

 

[Don]

This is precisely where I do disagree. Your use of the term "corrupted"
indicates that some damage has been caused to the scanned image.

If you edit the image in scanner software, you are changing raw data.
That is "corruption" of raw data.

Whether Photoshop (or anything else) is better than scanner software
or not is totally beside the point. Indeed, *what* software (doesn't
have to be Photoshop) is used to edit raw afterwards is totally beside
the point.

The subject and the point is, once again, getting the most, the
purest, the highest quality data out of the scanner. Stating that raw
"loses a lot of quality" is just completely wrong.

By you - I believe that Toby has qualified his statement specifically to
8-bit scans. However it is equally relevant to true 16-bit scans given
the known limitations of PS.

Again, you are changing the subject. This is *not* about Photoshop,
nor is it about 8-bit vs. 16-bit.

This is about scanning raw vs. using built-in scanner software image
editing tools, and then stating that raw "loses a lot of quality"!?

I am surprised that you are surprised. You have chosen to object to a
refinement of your inaccurate statement. I have never stated that you
were completely wrong on the matter, just not 100% correct in your
absolute terminology. "Exactly the opposite" workflow is not always
superior.

But you have failed to point out *where* I was "wrong".

All you do is interject "Analog Gain", or "Photoshop is not perfect",
etc. none of which has anything to do with comparison of raw vs.
cooked, and the assertion that raw "loses a lot of quality".

As I have made clear in several post throughout this thread, analogue
gain is just one aspect that happened to be the first process that
struck me as not being consistent with your statement that "exactly the
opposite" of Toby's was true but, as I mentioned very early on in the
thread, it is not a unique process that would be inconsistent with your
statement.

OK, be *very* specific and give me a *concrete* example of a "raw" vs.
"cooked" comparison where "raw loses a lot of quality".

And don't just make general statements, but give me a concrete,
detailed, specific example so we can nail down this exception you are
refering to.

Don.

 

[Kennedy McEwen]

I use the software but only to maximise
scan quality.

Isn't that what we all consider we do?

 

[Toby]

Excuse me for butting in, re: analog gain. Does the scanner not optimize
exposure by changing the analog gain based on the density of the target
film? Or am I again totally out to lunch here. When you speak about a raw
scan are you not using any sort of autoexposure controlled by the scanner
s/w based on the max transmission/density of the film?

Toby

 

[Bruce Graham]

Also, Don has observed that VueScan has a similar (but much worse)
effect on the scanner, so Don would - first of all, avoid using
VueScan - but should one start VueScan by mistake, it is strongly
advised to turn the scanner off before starting proper, non-beta
software such as NikonScan... ;o)

Just a data point with the Canon (which I realise no-one is talking about
I can happily switch back and forward between Filmget and Vuescan
without turning any hardware off.

 

[WD]

If you are scanning negatives, to get the max dynamic range from your
scans (without Nikonscan truncating highlights or shadows), beyond
Kennedy's suggestions, you can also scan as a positive. This will
capture oodles of dynamic range which you can trim in Nikonscan or
Photoshop. This way you can
sidestep Nikonscan's tendency to sometimes clip highlights or shadows

WD

 

[DavidTT]

As Don and I have stated repeatedly, the reason to get a raw scan is to
capture and preserve as much as information as possible at the scan
stage. For me, that means to get a raw scan by utilizing *all* the
hardware features available on a scanner that can improve the quality of
the scan, especially when what these hardware features can do cannot be
done by PS later. For a raw scan, I will only use the scanner sw to
control these hardware features, and nothing else. When I ask for a raw
scan on a Polaroid ss4000, the native sw will only scan at the maximum
12 bit depth and will disable all the corrections. Precisely what I
want. With other scanners' native sw, or with a third party sw, you may
need some more explicit directives to get a raw scan.

To clarify what I mean, I'm providing three separate lists. The first is
a list of basic hardware features found on a Polaroid ss4000, and should
be common on similar film scanners. The second is a list of model
specific hardware features. As far as I know, these are the exhaustive
hardware features that can influence a scan's quality, but I can stand
corrected. They are also functions which PS cannot do and therefore must
be done during scanning if desired. The third is a list of examples of
functions that are performed by scanner sw and can also be done in PS,
but are often mistaken to be done by hardware. BTW, when I choose a
scanner, I am most concerned about the hardware features in the first
two lists.

Basic hardware features on a Polaroid ss4000, and other similar film
scanners:

- cropping for the maximum scan area, i.e. 24x36mm in a full frame 35mm
slide mount
- default and fixed exposure
- default auto focusing
- 8 bit depth (also 12 bit depth on the ss4000)


Model specific hardware features on desktop film scanners:

- selection of 8, or more bit depths
- manual focus control
- exposure control, such as analog gain on Nikon (or long exposure?)
- hardware assisted noise reduction, such as Nikon's ICE
- multisampling in a single pass (very different from multi-pass
scanning)

Examples of functions performed by film scanner sw, and often mistaken
to be performed by hardware. To get a raw scan, you should disable all
these.

- film type selection
- tonal range (latitude) extension
- color correction
- sharpening

I welcome comments and corrections, especially from Don.

 

[WD]

- The scanner (software) will optimize, but it is not perfect.

- Yes, even with a 'raw' scan, there is still the issue of exposure whether
auto or manual. The raw scan allows capturing all the data as generated
by the A/D converter at whatever the given exposure is. The concept is that
in a non-raw scan, typically 'unused' dynamic range is tossed out. The
'raw' scan captures all the data.

Note in my previous post I discussed scanning a negative as a positive
in Nikonscan to capture a very large dynamic range. In my haste, I forgot
to mention of course that this approach would require inverting the scan
in Nikonscan or Photoshop. This is not difficult but somewhat tedious.
In addition, color balance can be tricky. I would not recommend this
as a normal workflow