Microtek 4000TF

[jbles]

Anyone had experience with this scanner? Am considering it because it seems
to be the only film scanner with profiles for b&w negatives which is
primarily what I need to scan.

Any help/advice appreciated.

jb

 

[Raphael Bustin]

Anyone had experience with this scanner? Am considering it because it seems
to be the only film scanner with profiles for b&w negatives which is
primarily what I need to scan.

Scanner profiles are useless, especially for
negatives (color or BW.)

With a bit of practice you'll be able to master
any decent film scanner.

Choosing a scanner based on available
"profiles" is ludicrous, IMO.

Mind you, I'm not saying the Microtek is
a bad scanner. For my money, right now,
I'd go with the new Minolta 5400 (assuming
all you want to scan is 35 mm.)


rafe b.
http://www.terrapinphoto.com

 

[jbles]

Why do you say scanner profiles are useless?

 

[Raphael Bustin]

Why do you say scanner profiles are useless?

Because, if the rest of your system (mainly
monitor and printer) are properly profiled,
then who needs 'em?

What you do need is to follow one of these
two workflows:

1. Scan "raw" and bring all 16 bits per channel
into your image editor, or

2. Take some care in the scanner driver to make
sure you have reasonable histograms with
approximately the gamma you want, no severe
color cast, and no data lost in the highlights or
shadows.

Having followed either of these procedures, all
that remains is to tweak the image in Photoshop
until it looks right on the screen.

Point I'm making is that it's not too hard to get
a scan that's good enough for any conceivable
purpose -- without a scanner profile. All it takes
is a bit of care in the scanner driver, or failing that,
a 16-bit scan that properly captures all of the
tones in the original.

The ICC standard really doesn't address the
handling of negative images anyway (refering
back to the original post which was concerned
with scanning BW negatives.)

 

[jbles]

Thanks for the explanation.

Because, if the rest of your system (mainly
monitor and printer) are properly profiled,
then who needs 'em?

What you do need is to follow one of these
two workflows:

1. Scan "raw" and bring all 16 bits per channel
into your image editor, or

2. Take some care in the scanner driver to make
sure you have reasonable histograms with
approximately the gamma you want, no severe
color cast, and no data lost in the highlights or
shadows.

Having followed either of these procedures, all
that remains is to tweak the image in Photoshop
until it looks right on the screen.

Point I'm making is that it's not too hard to get
a scan that's good enough for any conceivable
purpose -- without a scanner profile. All it takes
is a bit of care in the scanner driver, or failing that,
a 16-bit scan that properly captures all of the
tones in the original.

The ICC standard really doesn't address the
handling of negative images anyway (refering
back to the original post which was concerned
with scanning BW negatives.)

 

[Bart van der Wolf]

Because, if the rest of your system (mainly
monitor and printer) are properly profiled,
then who needs 'em?

I disagree.

Different scanners produce different raw data from the same piece of film.
The main difference is due to the spectral emission of the scanner
lightsource interacting with the particular spectral transmission of the
film dye-set. This will result is some colors to be of higher saturation on
one scanner, and lower on another.

The whole purpose of scanner profiling is to get a very similar response for
a certain dye-set regardless of the scanner used.

What you do need is to follow one of these
two workflows:

1. Scan "raw" and bring all 16 bits per channel
into your image editor, or

2. Take some care in the scanner driver to make
sure you have reasonable histograms with
approximately the gamma you want, no severe
color cast, and no data lost in the highlights or
shadows.

Having followed either of these procedures, all
that remains is to tweak the image in Photoshop
until it looks right on the screen.

Since no profile is assigned to the unprofiled scan, the colors will look
different in Photoshop. One could have assumed a certain colorspace, but
that is probably cause for more corrections to get a perceptually pleasing
image (which will still be different from the same scan on a different
scanner). So which one is close to the truth? It's a guess although both may
look pleasing, but different.

Point I'm making is that it's not too hard to get
a scan that's good enough for any conceivable
purpose -- without a scanner profile. All it takes
is a bit of care in the scanner driver, or failing that,
a 16-bit scan that properly captures all of the
tones in the original.

The ICC standard really doesn't address the
handling of negative images anyway (refering
back to the original post which was concerned
with scanning BW negatives.)

One can shoot a reflective IT8 target on negative film and use that to
create a profile with linear response in the density range of the target.
Without a profile, the shadows and higlights are very much compressed (toe
and shoulder of the characteristic curve) in different ways for the
different layers.

For Black and White film it can be beneficial to linearize the contrast, but
this can also be done in a post-processing step (based on a stepwedge or
exposure sequence).

Bart

 

[Raphael Bustin]

Since no profile is assigned to the unprofiled scan, the colors will look
different in Photoshop. One could have assumed a certain colorspace, but
that is probably cause for more corrections to get a perceptually pleasing
image (which will still be different from the same scan on a different
scanner). So which one is close to the truth? It's a guess although both may
look pleasing, but different.

So what does it matter that the two scanners look
different in Photoshop? If the rest of the system is
profiled and calibrated, you can edit the image in
Photoshop to look the way you want it to -- assuming
of course that the tonality of the scan is adequate
to begin with.

If you're shooting chromes for professional or
commercial purposes (eg. catalog photos) then
I would agree that a profiled scanner can be useful.

For personal use, or for shooting negatives,
"objectively accurate" color simply isn't a concern,
and in fact (when shooting BW or color negatives)
isn't even possible.


rafe b.
http://www.terrapinphoto.com

 

[Bart van der Wolf]

SNIP

So what does it matter that the two scanners look
different in Photoshop? If the rest of the system is
profiled and calibrated, you can edit the image in
Photoshop to look the way you want it to -- assuming
of course that the tonality of the scan is adequate
to begin with.

Unless you assign a profile to the image, profiling the rest of the system
is meaningless. At best you'll have a decent gamma approximation, but that's
all it is.

If you're shooting chromes for professional or
commercial purposes (eg. catalog photos) then
I would agree that a profiled scanner can be useful.

For personal use, or for shooting negatives,
"objectively accurate" color simply isn't a concern,
and in fact (when shooting BW or color negatives)
isn't even possible.

That may work (like with the Nikon scanners) but as I said, different
scanners result in different saturations similar to assigning sRGB,
ColorMatch or Adobe RGB to a file in your otherwise color profiled
environment. Basically the only difference is different assumptions for the
RGB primaries (probably all three wrong).

Bart

 

[Raphael Bustin]

SNIP

Unless you assign a profile to the image, profiling the rest of the system
is meaningless. At best you'll have a decent gamma approximation, but that's
all it is.

You can assign the profile to the image when you
open it in Photoshop. So what's the problem?

As to the rest of it being meaningless, I heartily
disagree. What matters to me is that my prints
match my monitor. And for that all I need is profiles
for the monitor and the printer.

There are lots of folks using ICC workflow without
profiling their scanners.

For that matter, I've been working in Photoshop 4,
without ICC profiles of any kind for years. If you pay
attention to your histograms and densitometer
readings, set your white points and black points
and neutrals by the numbers, you can kiss the ICC
bye-bye.

If you're scanning negatives, any assignment
to an ICC color space is entirely arbitrary anyway.

Digital imaging had a long rich history before
the ICC showed up to make everything all better...


rafe b.
http://www.terrapinphoto.com

 

[Bart van der Wolf]

You can assign the profile to the image when you
open it in Photoshop. So what's the problem?

Yes, but which profile?

As to the rest of it being meaningless, I heartily
disagree.

Guess we'll have to agree to disagree.

What matters to me is that my prints match my monitor.
And for that all I need is profiles for the monitor and the printer.

Yes, but that's the issue. If you have enabled (!) color management, Color
is converted on the fly. Without specifying what the profile (any profile)
of the scanner file is, you can be in for a surprise. As I said, try
assigning a (different) profile. Once you've assigned a/any profile, you can
rely on the screen reflecting the 'correct' color to balance, tonescale,
etc. But the saturation and linearity will not be accurate. It may look
pleasing, but it is not accurate.

There are lots of folks using ICC workflow without
profiling their scanners.

For that matter, I've been working in Photoshop 4,
without ICC profiles of any kind for years.

That was my first serious attempt at PS as well.

If you pay attention to your histograms and densitometer
readings, set your white points and black points
and neutrals by the numbers, you can kiss the ICC
bye-bye.

We disagree. The image will look neutral where expected, but was that red
tomato 150 or 200 in the red channel, and how much green and blue was there
to desaturate the color? The only way to come reasonably close from the
moment you open the file, is profiling. From there creativity kicks in.

If you're scanning negatives, any assignment
to an ICC color space is entirely arbitrary anyway.

After mask removal, preferably at scan time by balancing the RGB exposure
ratio, and inverting, it's the same as for slides (only with a larger scene
dynamic range).

Digital imaging had a long rich history before
the ICC showed up to make everything all better...

I know, been there, seen it, done it (almost) all.

Bart

 

[Raphael Bustin]

Yes, but which profile?

Whatever works.

I don't shoot photos for catalogs. I shoot land-
scapes. They need to please me first and
foremost, and if they do that they may end up
in my collection of saleable images.

I really don't put much stock in "objectively
accurate color." Heck, I'm not convinced that
it exists. For my prints, the color must be pleasing
and it must be believable.

"Memory colors" need to be accurate but
I can generally get them right without profiles.
Even skin tones are no real problem, for
the most part.

ICC profiling is way over-rated, IMO, with regards
to the needs of the average home user, or even
an advanced amateur.

Its main strength is making it easier to share
and compare -- eg. if I'm going to have my
files printed by a third party, or view my images
on someone else's monitor. Printer profiles
are useful for those who stray from OEM inks
and papers.

I just don't see a need for it at the point of image
creation, and I know I'm not alone in thinking that
way or going without them. (Though I have met
one or two who argue your side of this issue.)

There's a sidebar in Harald Johnson's book
(Mastering Digital Printing) about this very topic.

I've seen way too many people over the years
barking up the ICC color management tree, when
what they really needed was a bit of common
sense, and a bit of understanding of basic color
correction -- white points, black points, setting
neutrals, reading a histogram, working the Curves
tool, and so on.


rafe b.
http://www.terrapinphoto.com

 

[Neil Gould]

Hi,

Thanks for the explanation.

I hope that you take that explanation with a grain of salt, and read
Bart's excellent rebuttal of the notion that scanner profiles are
irrelevant.

The premise that any device will reproduce *the same* data from *the same*
film is simply false. The purpose of the profiling is to "normalize" the
result of a scan based on the characteristics of the film, not the image.
If the data is skewed at the time of the scan by virtue of an inadequate
or inaccurate interpretation of the film, further manipulation of the
image in an editor only degrades the results.

The other notion that I take issue with is that one can work solely from
the image on a monitor. You will require a rather sophisticated monitor (a
la one from the Barco line) and an environment with color controlled
lighting to work this way. Even then, there will be a learning curve to
understand how what one sees on the screen (transmissive color) relates to
what one sees in a print (reflective color).

If, as in Rafe's case, your work is "mainly artistic" -- i.e. there is no
objective reference involved -- it may not matter as much. However, if you
intend to match the color rendition of an original, the profiles can save
you a *lot* of time.

Regards,

 

[Raphael Bustin]

Hi,


I hope that you take that explanation with a grain of salt, and read
Bart's excellent rebuttal of the notion that scanner profiles are
irrelevant.

The premise that any device will reproduce *the same* data from *the same*
film is simply false. The purpose of the profiling is to "normalize" the
result of a scan based on the characteristics of the film, not the image.
If the data is skewed at the time of the scan by virtue of an inadequate
or inaccurate interpretation of the film, further manipulation of the
image in an editor only degrades the results.

So is it the scanner that you're normalizing (CCD + light source +
electronics) or the film, or the lighting of the original scene or
subject? IMO, most photographers wouldn't want to mess with
the last two, as that is part of the art and subjectivity of
photography, no? And of course there are other variables
involved -- eg, the optics of both the camera and the scanner
will affect the color and contrast of the scene as well.

Discerning photographers pick color films precisely because
of their particular interpretation of color. What color film can be
said to be "objectively accurate?"

The other notion that I take issue with is that one can work solely from
the image on a monitor. You will require a rather sophisticated monitor (a
la one from the Barco line) and an environment with color controlled
lighting to work this way. Even then, there will be a learning curve to
understand how what one sees on the screen (transmissive color) relates to
what one sees in a print (reflective color).

That is of course one of Dan Margulis' main complaints about
ICC color management -- over reliance on the appearance
of the image on a monitor.

I find that even when I work in ICC mode it's foolish and
dangerous to ignore the densitometer readings in Photoshop.

If, as in Rafe's case, your work is "mainly artistic" -- i.e. there is no
objective reference involved -- it may not matter as much. However, if you
intend to match the color rendition of an original, the profiles can save
you a *lot* of time.

I would posit that only a few professionals need to be concerned
about objectively accurate color. Indeed, if most photographers
were at all concerned about that, they wouldn't be shooting Velvia.


rafe b.
http://www.terrapinphoto.com

 

[Bart van der Wolf]

SNIP

So is it the scanner that you're normalizing (CCD + light source +
electronics) or the film,

Both. It is the spectral output of the lightsource that interacts with a
film's dye-set transmission.

or the lighting of the original scene or subject?

No, most definitely not. That's Tone- and Color balancing which comes after
normalizing the scanner response to the dye-set, and is at the discretion of
the artist.

IMO, most photographers wouldn't want to mess with
the last two, as that is part of the art and subjectivity of
photography, no?

I 100% agree. That's why we don't want the scanner to mess with them either.

And of course there are other variables
involved -- eg, the optics of both the camera and the scanner
will affect the color and contrast of the scene as well.

They remain essentially unaffected.

Discerning photographers pick color films precisely because
of their particular interpretation of color. What color film can be
said to be "objectively accurate?"

You're right they each have their characteristic, and it should not change.
All that profiling does is making sure that the characteristic is
maintained, across different scanners, thus eliminating one more variable.

The profiling makes sure that a color difference on film will have the same
relative difference in the scan. It makes no assumption (!) about the color
it should have been on film. The color on film is 'just' the result of
lighting, filters, lens, film characteristics and processing. Relative color
differences will be retained, but only after profiling.

Just to make it clear again. The IT8 target is not a simple exposure of a
number of color patches on film. Each(!) patch is exposed individually until
it's color is, with a small margin of error, the same as a reference. So
each film needs different exposures for each patch to approach the reference
values. Now you have a target that has exactly the same patch values as a
target on another film (but it assumes a certain viewing lightsource
quality, and the actual deviations from the reference are noted). Any
deviation *after* scanning from the noted input values must therefore have
been introduced by the scanner (lightsource/dyeset interaction), and the
profile can usually remove that influence.

Bart

 

[Neil Gould]

Hi,

So is it the scanner that you're normalizing (CCD + light source +
electronics) or the film, or the lighting of the original scene or
subject?

The lighting and subject are qualities of your image, and they aren't
factors of concern to a scanner profile. The other factors are components
of the profile, and affect how one scanner vs. another will interpret an
image.

IMO, most photographers wouldn't want to mess with
the last two, as that is part of the art and subjectivity of
photography, no?

Agreed. However, they shouldn't have been included as factors in scanner
profiling, per the above explanation.

And of course there are other variables
involved -- eg, the optics of both the camera and the scanner
will affect the color and contrast of the scene as well.

Again, parameters of the camera or scene are not relevant to the scanner
profile.

Discerning photographers pick color films precisely because
of their particular interpretation of color. What color film can be
said to be "objectively accurate?"

Of course, the selection of film is a decision that will affect the
rendering of a subject or scene. The "objective accuracy" that I referred
to has to do with how the film reproduces color, which is typically
regarded as a constant for a given batch. If that weren't reasonably true,
then color balance outcomes would always be unpredictable, and there
wouldn't be much point in chosing films based on their particular
interpretation of color. ;-)

That is of course one of Dan Margulis' main complaints about
ICC color management -- over reliance on the appearance
of the image on a monitor.

I find that even when I work in ICC mode it's foolish and
dangerous to ignore the densitometer readings in Photoshop.

Correct. Pros can do color correction on monochrome monitors, simply
because they go "by the numbers" rather than by what the monitor is
showing them.

I would posit that only a few professionals need to be concerned
about objectively accurate color. Indeed, if most photographers
were at all concerned about that, they wouldn't be shooting Velvia.

I'd tend to agree with you, as I don't shoot Velvia. ;-)

The other thing that I'd posit is that relatively few professional
photographers have had any training as professional scanner operators.
And, in this thread, that makes a difference w/r/t specific
recommendations. The OT suggested that profiles are important to him, and
I wouldn't presume that the reason is he (or other readers of this thread)
can't benefit from them.

Regards,

Neil

 

[Raphael Bustin]

Just to make it clear again. The IT8 target is not a simple exposure of a
number of color patches on film. Each(!) patch is exposed individually until
it's color is, with a small margin of error, the same as a reference. So
each film needs different exposures for each patch to approach the reference
values. Now you have a target that has exactly the same patch values as a
target on another film (but it assumes a certain viewing lightsource
quality, and the actual deviations from the reference are noted). Any
deviation *after* scanning from the noted input values must therefore have
been introduced by the scanner (lightsource/dyeset interaction), and the
profile can usually remove that influence.

Sorry Bart, you completely lost me here.

I've actually gone through the procedure of
profiling my Microtek 2500, using a 4x5" IT8
transparency target... one sheet of film.


rafe b.
http://www.terrapinphoto.com

 

[Bart van der Wolf]

In an attempt to clarify, I inserted some comments below.

This part describes (simplified) how an IT8 target (which you can buy) is
produced:
This part summarizes the target characteristics:
So, all targets have the same(*) patch colors when viewed under a 5000
Kelvin lightsource, and the deviations from the reference colors are noted
in the text file that goes with the target you bought. This will give you a
film with known colors that are close to a reference (which makes sure there
is a good spread throughout the gamut of the material).
(*) Some colors are at the extreme of the film used, so they may differ
between film types.

This part describes the evaluation of a scan of the target:
So, only the scanner influence is corrected, actual exposures have only the
color characteristics of the scene, camera lens/filter and film.

Additionally it is possible to take out the film characteristics by shooting
a relective target. Assuming the scanner is already profiled, the additional
influence of the film can be eliminated. That could be used for (color
negative) film which has to give a neutral (accurate) response (e.g.
reproduction of fabric colors, or more scientific applications).

Hope that helps.

Bart

 

[Neil Gould]

Hi,

Additionally it is possible to take out the film characteristics by
shooting a relective target. Assuming the scanner is already
profiled, the additional influence of the film can be eliminated.
That could be used for (color negative) film which has to give a
neutral (accurate) response (e.g. reproduction of fabric colors, or
more scientific applications).

Just to reinforce this point, that is the same procedure that I use for
shooting *either* negatives or slides. For each different lighting
condition of a scene, I shoot a color target as a reference and adjust the
scanning parameters accordingly.

Regards,