Is the EPSON 4990 really 16 bit?

[Oliver Kunze]

Hi,

after comparing 8 and 16 bit scans of 6x6 transparencies with the Epson 4990
and Epson scan 2.61G by thorough examination of the scans with photoshop
with respect to posterization in highlights and shadows, I can't see any
differences between the two bit depths. I scanned with Epson color
calibration as well as with ICM profiles.

Has anyone similar experiences?

Epson scan behaves a little bit strange with respect to 24/48 bit and color
profile handling. For example, when using ICM profiles, switching from 24 to
48 bit requires a relaunch of the exposure button and the resulting color
characteristics of the image is much different from the 24 bit version
(preview as well as the scan). Also, the tiff scan is not tagged with the
color profile, it must be assigned in photoshop after the scan.

Oliver

 

[Ken Weitzel]

Hi,

after comparing 8 and 16 bit scans of 6x6 transparencies with the Epson 4990
and Epson scan 2.61G by thorough examination of the scans with photoshop
with respect to posterization in highlights and shadows, I can't see any
differences between the two bit depths. I scanned with Epson color
calibration as well as with ICM profiles.

Has anyone similar experiences?

Epson scan behaves a little bit strange with respect to 24/48 bit and color
profile handling. For example, when using ICM profiles, switching from 24 to
48 bit requires a relaunch of the exposure button and the resulting color
characteristics of the image is much different from the 24 bit version
(preview as well as the scan). Also, the tiff scan is not tagged with the
color profile, it must be assigned in photoshop after the scan.

Hi Oliver...

You can't see 48 bit on your monitor

There are two reasons for scanning 48 bits as best I understand
it.

First, for archival purposes save all you can get, for who knows
what tomorrow may bring.

Second, scanning at 48 bit and doing your post scan tweaking at
48 bits will give you a much "smoother" histogram; that is to say
you won't end up with spikes.

Take care.

Ken

 

[Jim]

Hi Oliver...

You can't see 48 bit on your monitor

He means that you are looking at 24 bit images no matter what their source
had. You won't be able to tell the difference between 24 bit and 48 bit
images on a printer for the same reason.
Jim

 

[Robert Feinman]

Hi,

after comparing 8 and 16 bit scans of 6x6 transparencies with the Epson 4990
and Epson scan 2.61G by thorough examination of the scans with photoshop
with respect to posterization in highlights and shadows, I can't see any
differences between the two bit depths. I scanned with Epson color
calibration as well as with ICM profiles.

Has anyone similar experiences?

Epson scan behaves a little bit strange with respect to 24/48 bit and color
profile handling. For example, when using ICM profiles, switching from 24 to
48 bit requires a relaunch of the exposure button and the resulting color
characteristics of the image is much different from the 24 bit version
(preview as well as the scan). Also, the tiff scan is not tagged with the
color profile, it must be assigned in photoshop after the scan.

Oliver

You don't say which scanning software you are using, I'm assuming it
is the Epson scan software. If you do any adjustments using this
software it is possible that it does them in 8 bit mode and then outputs
in 16 bit. The Minolta software does this, for example.
16 bits doesn't change the end points just the number of steps in
between. So you may be using the auto features of the scan software
which will usually clip the ends to make the contrast higher. See
if you can adjust the end point manually.
Another thing is to try the shareware program Vuescan from hamrick.com.
This allows you to set all these factors yourself. Download it and
try it out.

 

[Oliver Kunze]

You don't say which scanning software you are using, I'm assuming it
is the Epson scan software. If you do any adjustments using this
software it is possible that it does them in 8 bit mode and then outputs
in 16 bit. The Minolta software does this, for example.
16 bits doesn't change the end points just the number of steps in
between. So you may be using the auto features of the scan software
which will usually clip the ends to make the contrast higher. See
if you can adjust the end point manually.
Another thing is to try the shareware program Vuescan from hamrick.com.
This allows you to set all these factors yourself. Download it and
try it out.
--

Robert,

I mentioned Epson Scan Version 2.61G. Meanwhile I downloaded version 2.68
from Epson's U.S. wesbsite but that didn't help. I nearly all the time
adjust the end points, since Epson scan's auto exposure usually clips the
highlights. My initual guess is that the 4990 hardware maybe not capable of
real 16 bit resolution, like the resolution promises which are usually not
fulfilled by flatbeds in this price range.

Oliver

 

[Sarah Brown]

Hi,

after comparing 8 and 16 bit scans of 6x6 transparencies with the Epson 4990
and Epson scan 2.61G by thorough examination of the scans with photoshop
with respect to posterization in highlights and shadows, I can't see any
differences between the two bit depths. I scanned with Epson color
calibration as well as with ICM profiles.

Has anyone similar experiences?

I've noticed that shadows are always posterised when I scan slides on my
4870, regardless of whether I used 16 bit mode or not. I suspect the scanner
just isn't up to the job, and that 16 bits with the 4870 (and it seems like
the 4990, from what you say) is just a waste of memnory and disk space. I
think I'm going to use 8 bit mode in future.

 

[Don]

He means that you are looking at 24 bit images no matter what their source
had. You won't be able to tell the difference between 24 bit and 48 bit
images on a printer for the same reason.

Actually, it's even worse than that. When it comes to color, our eyes
are really only around 6-bits per channel.

However, our eyes do have higher sensitivity to dynamic range (the
difference between light and dark) although we may need a minute or
two to adjust.

That's why High Definition Range (HDR) images with 32-bits per color
are meaningful. Even though we can't see all those colors we can
perceive the difference in brightness which all those extra bits
encode.

Of course, we'll have to wait for monitors to catch up. But it's a
good enough reason to archive 16-bit images and scan each image twice
(once for highlights and once for shadows) as those can later be
combined into an HDR image.

Of course, twin scans are also very useful to help combat noise even
if the end result is reduced to 16 or even 8 bits.

Don.

 

[Oliver Kunze]

I've noticed that shadows are always posterised when I scan slides on my
4870, regardless of whether I used 16 bit mode or not. I suspect the
scanner
just isn't up to the job, and that 16 bits with the 4870 (and it seems
like
the 4990, from what you say) is just a waste of memnory and disk space. I
think I'm going to use 8 bit mode in future.

The latter is the conclusion which I also have drawn. Which software do you
use with the 4870?

Oliver

 

[Sarah Brown]

The latter is the conclusion which I also have drawn. Which software do you
use with the 4870?

Epson Scan with Photoshop CS. I did try VS, even bought it, but found that
it was too unstable on my setup (PowerMac G4) to be useful (quite often it
wouldn't even launch - getting it to produce a preview was an achievement).

 

[Ronald Bruck]

Epson Scan with Photoshop CS. I did try VS, even bought it, but found that
it was too unstable on my setup (PowerMac G4) to be useful (quite often it
wouldn't even launch - getting it to produce a preview was an achievement).

Hmm. I've used VueScan with my 4870 on both a dual G4 and dual G5 and
had no problems. It has the annoying habit of refreshing every time
you drag the preview window, and when it does that, it SEEMS not to be
accepting input. I suppose if you can a large picture at a high dpi
you could see this.

Are you using OS X?

--Ron Bruck

 

[Sarah Brown]

Hmm. I've used VueScan with my 4870 on both a dual G4 and dual G5 and
had no problems. It has the annoying habit of refreshing every time
you drag the preview window, and when it does that, it SEEMS not to be
accepting input. I suppose if you can a large picture at a high dpi
you could see this.

Are you using OS X?

Yes. The machine in question is running 10.3. VueScan would occasionally
give me an image out, and when it worked it worked well, but I never did
manage to work out the set of circumstances under which it would work
reliably, so I gave up and went back to the Epson software.

 

[Kennedy McEwen]

Actually, it's even worse than that. When it comes to color, our eyes
are really only around 6-bits per channel.

6-bits evenly spaced throughout our perceptual range - which is about
the same as 13 bits linear encoding, as the output of a scanner CCD or
digital camera sensor would be.

 

[no one noteworthy]

--
----m0o0m

Yes. The machine in question is running 10.3. VueScan would occasionally
give me an image out, and when it worked it worked well, but I never did
manage to work out the set of circumstances under which it would work
reliably, so I gave up and went back to the Epson software.

I get the impression Hammrick's software is something of a cult thing. I
bought it too, for my 4870. I used it with Windows and couldn't achieve the
results everyone here seem to be able to do so I also returned to Epson scan
and have been quite happy with the scans.

My Nikon coolscan software is very unstable and I'm thinking of trying
vuescan with this but I think it definitely has some issues with Epson
scanners.

 

[Don]

I get the impression Hammrick's software is something of a cult thing.

That's a very polite way of putting it... ;o) Vuescan may be OK for a
quick-and-dirty web JPG but not for anything even remotely serious.

My Nikon coolscan software is very unstable

What specific problems do you have? Although NikonScan can be cranky
e.g. if you want to turn auto exposure off, in regular use it's very
reliable. Certainly several orders of magnitude more reliable than
Vuescan with its "bug du jour". ;o)

Don.

 

[Bart van der Wolf]

message SNIP

VueScan's author could have been of help there. Did you contact him?
There is a significant user base, on all sorts of platforms, so if
there is an issue with e.g. OS X, it's likely that he has already
heard about it from other users.

I get the impression Hammrick's software is something of a cult
thing. I bought it too, for my 4870. I used it with Windows and
couldn't achieve the results everyone here seem to be able to do

Why? You don't say what type of originals you were scanning, but it
should have been possible to get results similar to others ...

so I also returned to Epson scan and have been quite happy with the
scans.

Which is what ultimately counts, happy with the results.

My Nikon coolscan software is very unstable and I'm thinking of
trying vuescan with this but I think it definitely has some issues

Why would your Nikon software be unstable, and other users' versions
not? I don't hear other users complaining, so I must assume they are
fine. Could it be something specific about your setup? Bad cable? Just
guessing, because you're not being very specific.

with Epson scanners.

Again, I also don't hear about scanner stability issues from other
Epson users. I also have an Epson, and a Nikon, and a Minolta
connected to my PC all driven by VueScan (I love the single interface
for all).

Bart

 

[no... it's not!]

Why? You don't say what type of originals you were scanning, but it should
have been possible to get results similar to others ...


Which is what ultimately counts, happy with the results.


Why would your Nikon software be unstable, and other users' versions not?
I don't hear other users complaining, so I must assume they are fine.
Could it be something specific about your setup? Bad cable? Just guessing,
because you're not being very specific.


Again, I also don't hear about scanner stability issues from other Epson
users. I also have an Epson, and a Nikon, and a Minolta connected to my PC
all driven by VueScan (I love the single interface for all).

Bart

History Bart, tells me you are more comfortable with software that needs
fiddling than out of the box, functionality that just works.
..
The developer of vuescan asked me to participate (supplying logs etc) in his
development of the 4870 interface when I had problems getting the scanner to
work at all. At the time, many people in this group reported out of memory
problems with the (then very new) 4870 and shortly after, Epson released a
new driver which did little to address the real problem.

The problem was all about memory allocation and subsequently fixed by
altering the Windows boot.ini file and knocking back the amount of RAM
Photoshop was using as well as setting the swap file to a fixed size.

In the mean time I had bought vuescan. I couldn't obtain a colour balance
when scanning Fuji ISO 200 negative film. I could not easily get a colour
balance when scanning Kodak ISO 400 negative film either and the solution of
developing a "profile" for a bought application just to make it work, is
distasteful to me. If you are going to sell something that need further
development before it works, you really need to look closely at your
business ethics.

The Nikon software crashes if left open. When you return to load a fresh
stack of slides, the system may reboot itself when you hit the preview
button or it may just produce the unstable message - do you want to tell
Microsoft about the problem, or it may just do as it's designed to.

It seems that there is some memory flooding going on and Nikonscan is the
culprit. Available RAM reduces with the progress of time (hours of scanning)
when Nikonscan is running. The problem may be the VIA chipset on the board
of this PC but if that were true, why then would all the other software
perform correctly?

Removal and reinstallation is a waste of time. The workaround currently is
to simply restart the program and continue on. With negatives, the problem
is often sever enough to require a reboot of the system. This is not a "scan
an occasional film" shop. We run this scanner at least 7 hours a day. If I
were to just scan a film here and there, it would have no problem but this
is not what I bought the scanner for and it should be able to work
continuously. It does not have a "duty cycle" to limit it's work time.

 

[Sarah Brown]

VueScan's author could have been of help there. Did you contact him?

No, I wasn't sufficiently unhappy with Epson scan, and found myself a little
too daunted by the way to make VueScan work with colour-profiles, to
consider it to be worth the bother.

I'm not trashing VueScan here - I know that it's generally well regarded,
and I appreciate Ed's contribution. I just found that what came with the
scanner was mostly "good enough" for my purposes, given the difficulty I
was having getting VueScan working.

 

[Paul]

Hi,

after comparing 8 and 16 bit scans of 6x6 transparencies with the Epson 4990
and Epson scan 2.61G by thorough examination of the scans with photoshop
with respect to posterization in highlights and shadows, I can't see any
differences between the two bit depths. I scanned with Epson color
calibration as well as with ICM profiles.

It sounds as if you and others in this thread carry traditional
misunderstandings of what additional bit depth is for, and what is to
be gained.

The truth is that you should NOT expect to see any differences between
8 and 16 bit scans of the same original from the same scan settings.
Both 8 and 16 bits are enough to render a smooth image without
posterization. The high-end prepress industry worked with 8-bit images
for years (and the majority probably still does). If you see
posterization, one or more of the following things has happened. The
scanner tone curve and/or end points are set incorrectly or too
extreme, stretching out available levels to the point where they are
posterized. Or the scan is fine but your monitor is not properly
profiled, so that you are perceiving posterization where there isn't
any. Or the scanner has a bad signal-to-noise ratio in that portion of
the tonal range.

As far as that side discussion about the human eye, it isn't really
applicable to the problem at hand. That's not what the high bit depth
is for, otherwise 6-bit scanners would be fine. The additional bit
depth is there for editing headroom. The extra bits help postpone
posterization as you shift levels around during editing. In accordance
with the previous paragraph, if it's possible to set the scanner
software so that the resulting 8-bit scan is optimized well to the
tonal range, and you get a perfect scan to begin with, then there is no
need to edit and no need for additional bits beyond what the human eye
can perceive.

But if a scan needs editing, the extra bits provide "wiggle room" so
that image levels that are pushed around during editing aren't simply
clipped to meaningless posterized blobs. The closer your original scan
is to being perfect, the less 16 bits is going to help. Flipping that
over, the more your original needs heavy tonal moves, the more it will
be preserved by going to 16 bits. Another corollary to this is that the
better you are at image editing, the fewer number of corrections you
need (like golf), and the fewer bits you need to scan at because your
list of cumulatively damaging changes is going to be shorter than a
novice's. If I think I can correct a scan in one good move, I'll scan
at 8 bits and save the disk space.

Also, you are limited by what the original film or print contains. Even
if you scan at 16 bits, if you apply a tone curve that stretches an
area of an image beyond the number of levels available in the original,
it's going to posterize.

Realize that it is normal for an 8-bit and 16-bit image to look
completely identical. You won't see the difference until you start
editing repeatedly and the damage from changes accumulate.

 

[Sarah Brown]

It sounds as if you and others in this thread carry traditional
misunderstandings of what additional bit depth is for, and what is to
be gained.

The truth is that you should NOT expect to see any differences between
8 and 16 bit scans of the same original from the same scan settings.

My problem with the 4870 is that quite often I'm not seeing differences
between an 8 bit scan, and that same scan posterised to 6 bits. I therefore
conclude that operating the scanner in 16 bit mode is largely a waste of
time and memory.

Both 8 and 16 bits are enough to render a smooth image without
posterization. The high-end prepress industry worked with 8-bit images
for years (and the majority probably still does). If you see
posterization, one or more of the following things has happened. The
scanner tone curve and/or end points are set incorrectly or too
extreme,

I set the end points to where they need to be to avoid "blocking out"
highlights (particularly gross on the 4870 if you get this wrong) and
shadows. Where those are is entirely down to the film type used and the
photograph taken.

Realize that it is normal for an 8-bit and 16-bit image to look
completely identical.

It's not only normal, it's mandatory for most of us, since we're using
graphics hardware which is limited to 8 bits per channel.

 

[Oliver Kunze]

----- Original Message -----
From: "Paul" <[email protected]>
Newsgroups: comp.periphs.scanners
Sent: Thursday, January 26, 2006 12:12 AM
Subject: Re: Is the EPSON 4990 really 16 bit?

It sounds as if you and others in this thread carry traditional
misunderstandings of what additional bit depth is for, and what is to
be gained.

Paul,

I think there have been misunderstandings due to my initial posting. I did
not mean that I want to see a difference between 8 and 16 bit by simply
looking at unedited raw scans on the screen. I meant that there is no
difference between 4990 8 or 16 bit scans even after sophisticating tonal
correction in cases when a wider bit-depth should bring benefits. I scan 6x6
transparencies and as Sarah Brown has already mentioned scans of these
images who include the complete tonal range of the film image usually
require (dependent from the contrast of the image) a certain amount of
curve and/or level corrections in order to achieve a good result suitable
for screen or print output with good overall contrast.

My question aimed at the 16-bit capability of the Epson 4990 flatbed scanner
and if anyone made the same or different observation than me, that image
acquisition by this scanner has only 8-bit quality.

Oliver