Mustek 1200 III EP

[Dave]

A few weeks back I posted an article about problem(s) I was having with the
above scanner. Not only do I have it working, I'm starting to like it even
though it's slow. I have read through the information (well, some of it) on
www.scantips.com. What a resource! My grainy, off-color scans of old
photos have been completely revitalized with the use of editing tools in an
old version of PhotoImpact.

Now, one of the tips on the scantips page said that filters like Unsharp
Mask work BETTER if done at the scan level as they can work on the 30 or
36-bit image data as opposed to some degraded TIF or JPG or PNG version of
the same data which has been saved and reopened. The Mustek driver for my
scanner HAS an Unsharp Mask filter, but there are no adjustments (radius,
amount, threshold). I guess Mustek figures their defaults are good for
everyone or, more likely, that most people who will buy the scanner will
find it good enough. It's not BAD, but performs poorly on larger scans
which could be entirely corrected if one could adjust the parameters. I
note that whatever program I use to do the actual scanning (I have tried
half a dozen or so) uses the Mustek TWAIN interface, part of which is the
limited filtering. One other thing I find lacking is Mustek's histogram
tool. It's a histogram tool in name only, and I find it completely useless.

My questions are, specifically: is it true what the guy on scantips.com
says regarding filtering the raw scan? Or can I save the scan image in a
format which retains ALL of the original scan data? I am guessing there is
because JPG images from my digital camera show "RGB True Color (24-bit)" in
the properties. Also, is there any other driver I can use for this scanner
which might have more advanced filtering abilities?

Thanks for any and all replies.

Dave

 

[CSM1]

A few weeks back I posted an article about problem(s) I was having with the
above scanner. Not only do I have it working, I'm starting to like it
even
though it's slow. I have read through the information (well, some of it)
on
www.scantips.com. What a resource! My grainy, off-color scans of old
photos have been completely revitalized with the use of editing tools in
an
old version of PhotoImpact.

Now, one of the tips on the scantips page said that filters like Unsharp
Mask work BETTER if done at the scan level as they can work on the 30 or
36-bit image data as opposed to some degraded TIF or JPG or PNG version of
the same data which has been saved and reopened. The Mustek driver for my
scanner HAS an Unsharp Mask filter, but there are no adjustments (radius,
amount, threshold). I guess Mustek figures their defaults are good for
everyone or, more likely, that most people who will buy the scanner will
find it good enough. It's not BAD, but performs poorly on larger scans
which could be entirely corrected if one could adjust the parameters. I
note that whatever program I use to do the actual scanning (I have tried
half a dozen or so) uses the Mustek TWAIN interface, part of which is the
limited filtering. One other thing I find lacking is Mustek's histogram
tool. It's a histogram tool in name only, and I find it completely
useless.

My questions are, specifically: is it true what the guy on scantips.com
says regarding filtering the raw scan? Or can I save the scan image in a
format which retains ALL of the original scan data? I am guessing there
is
because JPG images from my digital camera show "RGB True Color (24-bit)"
in
the properties. Also, is there any other driver I can use for this
scanner
which might have more advanced filtering abilities?

Thanks for any and all replies.

Dave

The scantips guy knows his stuff, but he is human and sometimes makes
mistakes.

Due to the age of that scanner, the manufacturers driver is probably the
best available.

The most you can do is use the best optical resolution the scanner has to
offer, use/not use unsharp mask as the preview shows you. If there are
exposure controls, use them to get the best scan.
Histograms are a fairly new concept in scanner tools. A lot of older
scanners do not have that function.

Saving the scan as a TIFF will preserve the maximum data the scanner
provides externally. If you can save a 16 bit TIFF from that scanner you
will have the most that you can save. 16 bit tiffs are sometimes huge files.

 

[Dave]

Due to the age of that scanner, the manufacturers driver is probably the

best available.

I had assumed as much, I think the scanner came out in 1997.

The most you can do is use the best optical resolution the scanner has to
offer, use/not use unsharp mask as the preview shows you. If there are
exposure controls, use them to get the best scan.
Histograms are a fairly new concept in scanner tools. A lot of older
scanners do not have that function.

That's another problem with the Mustek driver: what you see in the preview
DOES NOT accurately reflect what you will see in photoshop or photoimpact.
It's too bad because the preview always looks better than what you get as a
final result.

Saving the scan as a TIFF will preserve the maximum data the scanner
provides externally. If you can save a 16 bit TIFF from that scanner you
will have the most that you can save. 16 bit tiffs are sometimes huge files.

Yeah, I noticed that. A 4x5" scan at 300 dpi is 3.5-6Mb. Thank goodness
for enormous inexpensive hard drives. Is there any quality difference
between "TIF uncompressed", "TIF with LZW compression" and "TIF with
packbits compression"?

Dave

 

[Dave]

use/not use unsharp mask as the preview shows you. If there are

exposure controls, use them to get the best scan.

Would I be better off to use the scan driver's adjustments or just scan it
with no adjustments and use Photoshop to do the messin'?

 

[CSM1]

use/not use unsharp mask as the preview shows you. If there are

Would I be better off to use the scan driver's adjustments or just scan it
with no adjustments and use Photoshop to do the messin'?

I personally think that adjustments in the scanner driver are better than
post adjustment.

The reason I think that is because the scanner driver will actually tell the
scanner hardware to perform the adjustments such as adjustment of the
exposure, which makes a huge difference in the final output.

One of the most important adjustments to the scanner is the black and white
points (exposure). If the white point setting clips the whites of the
document (picture), you lose the detail in the highlights. If the black
point is set too high, you lose the shadow details. These can not be
recovered in a post process, you can not put back what was lost in the
original scan.

As for the preview and the final scan not appearing the same, you may not be
using color space management.

Your monitor and the scanner are not using the same color space.

The whole system is monitor, scanner and printer creating the same
appearance of color.
Scan a picture, see that it looks good on the monitor, and print on the
printer and it also looks good. People look like people with the same color
of skin on all three.

 

[Wayne]

Now, one of the tips on the scantips page said that filters like Unsharp
Mask work BETTER if done at the scan level as they can work on the 30 or
36-bit image data as opposed to some degraded TIF or JPG or PNG version of
the same data which has been saved and reopened.

Oops! Sorry Dave, I better go check that page to see where it gave you the
wrong idea. It does show a USM tool in a scanner driver (maybe I better
hide that), but my site does NOT say "filters like Unsharp Mask work
BETTER if done at the scan level as they can work on the 30 or 36-bit image
data". Unsharp Mask is not a tone shifting operation, so it doesnt need
more than 24 bits.

You probably read something like that about the histogram tool, but the
only operation that really needs more than 24 bits is gamma (already
automatically done unless you can output RAW). And (if assuming cases of
drastic data shifts) it might be argued for the histogram/curve tools.
But not USM or other filters.

Regarding the USM page: http://www.scantips.com/simple6.html

The last sentence of the second paragraph says:

And perhaps there will be additional processing in the image program, and
the sharpening really should be done as last step.

The last sentence on the page says:

USM should be the final operation, after all other adjustments

The meaning (altho maybe not said clearly) is to get the image into a good
editor, finish ALL of your editing and resampling that might be needed, and
then do USM LAST (implying not done in the scanner). At least that is my
own opinion.

 

[Dave]

Oops! Sorry Dave,

Naw, you are right.

It was a huge volume of info to digest for a novice scanner. Unfortunately
my scanner driver is old and lacks a histogram tool. I do find that if I
have, say, a dark picture (all of my old Polaroids seem to have darkened
with age) that it does work better to use the "brightness" control on the
scan driver than to do nothing at all. I can't adjust the histogram tool in
my editing program enough with a "flat" scan on these photos to bring back
the colors and detail in the dark dark shadows. I will also note that I am
having a really hard time with USM in PhotoImpact. Mostly what I want to
sharpen are faces and I have spent about an hour screwing with just one
photo. USM seems to sharpen everything else FIRST, then there is a
threshold which, when crossed, suddenly turns everyone's teeth electric
white and gives them a halo around their light-colored clothes and often
makes discolored patches on their faces. A lot of the photos I am trying to
fix up are mid-70's color photos with matte finishes. They have either
yellowed somewhat with age or perhaps they were always that way, but USM's
addition of white (and it always seems to add white) is just glaringly
obvious given the complete lack of any white in the original photos. I
guess I'll just have to keep at it or maybe just accept that I can't
"create" detail that isn't there in the original photos. I guess I could
use the histogram too to turn yellow back into white... hmmm....

Thanks for all of your efforts, I think you shortened my learning curve by
about 5 years. Unfortunately I'm still on the steep part.

Dave

 

[Don]

I personally think that adjustments in the scanner driver are better than
post adjustment.

Actually, exactly the opposite is more likely to be true, because...

The reason I think that is because the scanner driver will actually tell the
scanner hardware to perform the adjustments such as adjustment of the
exposure, which makes a huge difference in the final output.

.... that relies on the scanner software "guessing" and the operator
agreeing by looking through the preview "keyhole".

If one cares for such high level of quality the best is to do as
little as possible in the scanner software, limiting the settings only
to hardware related options which can not be done later in
post-processing. Otherwise known as "scanning raw" - my personal
mantra... ;o)

But anything that can be done later is certainly better done in a
dedicated image processing software with the full complement of tools
and with all the ability to examine the image at maximum resolution
and magnification, using the full 16-bit histogram, etc. None of which
is normally available in the scanner software (nor should it be!).

One of the most important adjustments to the scanner is the black and white
points (exposure). If the white point setting clips the whites of the
document (picture), you lose the detail in the highlights. If the black
point is set too high, you lose the shadow details. These can not be
recovered in a post process, you can not put back what was lost in the
original scan.

And it's the scanner software that's losing it!!! All you're doing
with black and white point is telling the scanner software to do this
*after* the scan has already been completed and before the image is
transferred.

Just like everything else, that's much better done in an image editing
program afterwards. Not only can you experiment with different B&W
point settings *without* any loss (or need to re-scan) because you can
always start with the "raw" image again, but - and this is very (!)
important - the B&W point you set in scanner software is based on the
histogram of the *preview*!!! That's a *fraction* of the actual image!

By processing the image in an image editing program afterwards you'll
be able to see the *real* histogram of the actual data, rather than
the *guesstimate* of the scanner program based on a low resolution
preview scan!

If maximum quality and purest data is the goal, the scanner software
should only do what it's supposed to do - scan.

Don.

 

[Wayne]

Naw, you are right.

It was a huge volume of info to digest for a novice scanner. Unfortunately
my scanner driver is old and lacks a histogram tool. I do find that if I
have, say, a dark picture (all of my old Polaroids seem to have darkened
with age) that it does work better to use the "brightness" control on the
scan driver than to do nothing at all. I can't adjust the histogram tool in
my editing program enough with a "flat" scan on these photos to bring back
the colors and detail in the dark dark shadows. I will also note that I am
having a really hard time with USM in PhotoImpact. Mostly what I want to
sharpen are faces and I have spent about an hour screwing with just one
photo. USM seems to sharpen everything else FIRST, then there is a
threshold which, when crossed, suddenly turns everyone's teeth electric
white and gives them a halo around their light-colored clothes and often
makes discolored patches on their faces. A lot of the photos I am trying to
fix up are mid-70's color photos with matte finishes. They have either
yellowed somewhat with age or perhaps they were always that way, but USM's
addition of white (and it always seems to add white) is just glaringly
obvious given the complete lack of any white in the original photos. I
guess I'll just have to keep at it or maybe just accept that I can't
"create" detail that isn't there in the original photos. I guess I could
use the histogram too to turn yellow back into white... hmmm....

Thanks for all of your efforts, I think you shortened my learning curve by
about 5 years. Unfortunately I'm still on the steep part.

Dave

I am pretty guilty, because that web page about USM had not had any attention
in a very long time, and I was a bit surprised what it still said, and didnt
say. I do continually work on the book version, but tend to forget about
what the web version might say. But I have just now updated that USM web
page, and it should be more clear now.

The brightness control possibly may be your "white" problem. Brightness
controls simply add a constant value to all image tonal values (without
intelligence is my way of thinking about it). It is an older tool, from the
days when computers couldnt do nearly as much. For example, if the magnitude
of the brightness setting is so that it is adding a value of 40,
it adds 40 to 0 to get 40 (which reduces possible contrast at black end)
it adds 40 to 130 to get 170 (which is the brightness seeked)
it adds 40 to 220 to get ... oops, it can only clip (truncate) at 255, so
many light values become featureless full 255 (loss of bright detail). The
final histogram will show a large spike right at 255, which is rarely good.

I am assuming results are a full histogram (data pretty much extending to
both ends of range, but without large spikes at ends), so adjust the Black
and White end points first to achieve that. That sets contrast. Then if you
want to increase image brightness, so that the image views brighter, then
increasing (sliding left, increasing number) the midpoint of a histogram tool
is a much better way. Or raising the midpoint of the curve tool is
essentially the same thing. Both are just a bit higher value of gamma.
Neither of these adjustments much affect the White Point, so it doesnt clip
at white. Either of these will leave some dark tones at 0, retaining black.
Overall contrast is retained. The image is simply brightened, in a very
pleasing way. Doing histogram adjustments in the scanner driver is good, but
later in an editor program is also OK. Scanners without this histogram tool
are doing some of this automatically anyway, so the later editor adjustment
is probably a modest adjustment, not a drastic adjustment.

Versions of PhotoImpact earlier than 8 had an extremely aggressive USM, and
the default settings need to be very greatly reduced. It sounds like you may
be overdoing it quite a bit. General advice is: If it causes it to look bad,
then dont do that

 

[Alan Browne]

Thanks for any and all replies.

In passing only ... I've had the same scanner for over 5 years. I don't
use it much as I scan from negative/slides most of the time. Sometimes
I use it for printed materials or when sombody has a photo to restore.

Recently I'm getting glitchy scans (a line across the image, wrong
colors) and it can take a couple tries to get it right. The culprit is
probably me as I go on with other CPU heavy tasks during the scan.

My main "nit" with the Mustek driver is that it puts up a nonmoveable,
stay-on-top "cancel" dialog box in the middle of the screen ... possibly
to stop people from using the CPU for other tasks... that is to say, a
pretty badly written driver in the first place. But while it is there,
it makes other tasks a pain.

I found and installed the WinXP version (which If I recall is a Win2K
version patched for something else...) so at least it's alive.

So, I'll watch this thread in the vain hope that a better driver gets
linked.

Cheers,
Alan