Epson 4490 Scanner - Tests and Settings.

[Harry Stottle]

It's normal to use gamma 2.2 on Mac also these days...

The default gamma setting on the 4490 seems to be 1.8, and I had
presumed this had been set for Macs. If the default has been set to 1.8,
when it should be 2.2, this could partly explain why some people are
struggling to get good results.

PS: Your site throwed malware at me, check your advertising
conncetion. ds

Sorry, I did a search for free web hosting and those sites came up. I
tried the first one 3 times before posting the link, and didn't get any
pop ups, but as soon as I posted the link here I got them. I wanted to
place the info on a free site so it could be left there indefinitely,
and anyone searching the newsgroups in future could locate the
information, even months ahead. Hopefully, the advantages will outweigh
the disadvantages if someone needs help setting up the scanner, and if
someone wants to put the information, including the images, on a site
without pop ups, they have my permission to do this.

 

[Peter Chant]

I have said repeatedly that the 4990 is a good value. From
what I've seen, it's roughly on par with the 2700 dpi film
scanners from about 5 years ago. I paid $1350 for one
of those, way back then... and it could only scan 35 mm.

I'm not sure it is that good on 35mm, I'm very tempted to get my Acer
Scanwitt up and running again. IT is clear that the Epson has the better
dmax but the Acer is definitely sharper.

 

[Raphael Bustin]

I'm not sure it is that good on 35mm, I'm very tempted to get my Acer
Scanwitt up and running again. IT is clear that the Epson has the better
dmax but the Acer is definitely sharper.

Like I say: "roughly on par" with a Polaroid Sprintscan
35+, vintage 2000 or so, rated 2700 dpi optical. That
is perhaps giving the Epson the benefit of the doubt.

These days I would expect more for a 35 mm film
scanner. OTOH, I'm not scanning much 35 mm film
any more, except for old archives. I use my 4990
for LF, mostly for want of a better alternative.


rafe b
www.terrapinphoto.com

 

[if]

You'll find a few such comparisons on my scan-snippets site.
(Epson 4990 vs. Coolscan 8000.)

The 4490 should give the same resolution and image
quality; its main limitation will be max the scan area.

I am currently looking at getting an epson scanner, according to the specs
the 4490 is Dmax 3.4, whilst the 4990 is Dmax 4.0. Can anyone say if this
makes much difference in practice? I have heard that Dmax figures are
rather theoretical as usability of the denser details also depends on
things like CCD noise. Also, I hear that only Velvia has a Dmax of more
than 3.5 (I have some Velvia stuff but not that much).

I currently have an old Minolta Dimage Scan Dual but rarely use it because
removing dust spots is a big pain (no ICE) and it can only scan one slide
at a time. Also some play in the holder tends to mean images having to be
rotated a bit to straighten them, losing quality.

The Epsons appeal to me because of the ICE and possibility of scanning 8
mounted slides / 24 unmounted at a time (4 / 12 in the case of 4490). Is
this a realistic prospect? Can I set it up to save them all to disk
unattended (I don't have enough RAM to hold several scans in memory at
once)?

With the 4990 being twice the price of the 4490, I wouldn't want to shell
out for the 4990 if I couldn't be sure of making use of the extra scanning
capacity or getting better results. OTOH I don't want to be in the position
of feeling a need to rescan my images on something better at a later date!

 

[Raphael Bustin]

I am currently looking at getting an epson scanner, according to the specs
the 4490 is Dmax 3.4, whilst the 4990 is Dmax 4.0. Can anyone say if this
makes much difference in practice? I have heard that Dmax figures are
rather theoretical as usability of the denser details also depends on
things like CCD noise. Also, I hear that only Velvia has a Dmax of more
than 3.5 (I have some Velvia stuff but not that much).

I wasn't aware of the difference in Dmax
rating between the two. Sorry.

If you scan mostly negatives or documents,
a Dmax of 3.4 should be enough.

For chromes, the extra Dmax could help.

Dmax for most C41 films is in the mid-2 range
(log scale) and for chromes can approach or
exceed 4.

Whether Epson's numbers reflect reality is
another matter.


rafe b
www.terrapinphoto.com

 

[Noons]

Whether Epson's numbers reflect reality is
another matter.

they don't...

 

[Raphael Bustin]

they don't...

Actually, the DR of the Epson isn't that bad.

It does a vaguely respectable job on the scan-comparison
test involving Neil Snape's very dense Fuji chrome.

(See the snippets page, about halfway down..)


rafe b
www.terrapinphoto.com

 

[Harry Stottle]

I have used your settings for color positive film with success.
Have you started to scan black and white film yet. Now I am playing
with black and white and grey scale learning the ropes.

I have just been playing around with 35mm B&W negatives, and these are
the settings I have been using.

Mode - Professional

Document Type - Film

Film Type B&W Negative Film

Image Type - 16 bit Greyscale (or 8 bit)

Scanning Quality - Best

Resolution 4800 (or 3200)

Do a 'Preview' and select image with the marquee tool.

In 'Configuration' select 'Color' tab and tick 'No Color Correction'

These settings have given very good results, and required minimal
further adjustments.

If anyone knows of any free webspace, without complications, let me know
and I will post an example B&W image..

 

[Guest]

Thanks Harry for taking the time to share your procedures.
Denis

 

[Noons]

Actually, the DR of the Epson isn't that bad.

It does a vaguely respectable job on the scan-comparison
test involving Neil Snape's very dense Fuji chrome.

(See the snippets page, about halfway down..)

Yeah, I know what you mean. But try as I might,
I can't get it to show the eyes here:
http://members.iinet.net.au/~nsouto/photos/2006-10-02.jpg

That's 3200dpi rez vuescan, treated for grain in 48-bit raw
with NI, vuescan again into a 24-bit tiff with gamma
adjust, sharpened and jpg'ed with gimp as a final output. I
sharpen it more for a print, this is enough for a monitor.

As you can see, it's reasonably tack for the class of scanner
we're talking about and there is no sign of the obnoxious grain
"effects" I referred to when talking about scanners previously.
In particular, the oof background is "clean" as opposed to horrible
grain aliasing that happens otherwise.

Yet, the eyes are mostly gone. In the film: certainly visible with
a strong loupe. I know this is Velvia, high contrast, yaddayadda.
But if I can distinguish the eyes fine in the original, so should the
scanner.

Not enough DR or some scan error on my part, take your pick.
But I reckon I won't get better results than this with the 4990.
Roll-on the ls9K whenever Nikon Australia gets its act together...

 

[Raphael Bustin]

Not enough DR or some scan error on my part, take your pick.
But I reckon I won't get better results than this with the 4990.
Roll-on the ls9K whenever Nikon Australia gets its act together...

Be interested to see the comparison when you get your
Nikon and learn to use it. Be interested to get your
reaction to the grain also, when that happens.

The reason this image is smooth (well, one of the
reasons) is that it's not coming close to resolving
the grain. I have a hunch that when you get your
9000, you're going to want a copy of NeatImage
with that.

Some folks hate the sight of grain. Me, I've always
taken it for granted. It's an old acqaintance, at least.


rafe b
www.terrapinphoto.com

 

[Noons]

The reason this image is smooth (well, one of the
reasons) is that it's not coming close to resolving
the grain. I have a hunch that when you get your
9000, you're going to want a copy of NeatImage
with that.

Sorry, not quite. This is what I've tried to explain before
and haven't been able to. The scanner resolved a LOT
of the grain in this film image. Look closer. So much in fact
that it went into grain aliasing, a typical event with the
current crop of scanners. That, was the problem.
Which I've found a solution to.

This:
http://members.iinet.net.au/~nsouto/photos/2006-10-03.jpg
is EXACTLY and precisely the same scanned raw file out
of vuescan. I processed it into a 24-bit tiff with vuescan,
then sharpened with gimp with EXACTLY the same settings
as the first image. Indeed, the grain is there. So is the ugly
grain aliasing. Which is NOT in the original film frame, no
matter how strong a loupe I use.

This:
http://members.iinet.net.au/~nsouto/photos/2006-10-02.jpg
is EXACTLY the same raw scanned file, but with the
grain aliasing removed. I did this by processing the raw
48-bit scanned file through NI and THEN vuescan it into
24-bit tiff and sharpened with the same settings as the other.
Ie, same workflow, same adjustments, only diff is I did filter
out the grain alias BEFORE I converted from 48-bit raw scan.

If you look closely, the grain is there allright in
http://members.iinet.net.au/~nsouto/photos/2006-10-02.jpg
But it's now the true grain of the velvia emulsion, the so-called
"high-frequency" component of the NI folks. Which I don't
filter: I like it as it is!

That ugly "clumped" stuff in the oof areas of the first image
which is NOT in the film is nearly completely gone. This
is the "grain" that I object to in most scanned example images
I see in your pages and others: I know there is no way that
clumpy grain is in the original film, it's an introduced scanning
artifact. Assuming of course a modern 100ISO E6 film or K25!

Some folks hate the sight of grain. Me, I've always
taken it for granted. It's an old acqaintance, at least.

Same here. Grain aliasing is however something
that I've had to live with since scanning. It was
never there and I never saw it with enlargers.
This is the demonstration I intended to show you
folks before. This time done within the exclusive
realm of scanned images so it's easier to compare.
Both images are near equally sharp in the focused areas,
as should be expected: they come from the same
raw scan file! One shows the very fine grain in oof
areas expected out of a film like Velvia. The other is
the basic ugly grain aliasing that is the result of
scanning at a rez near the grain size.

It still doesn't resolve my major leftover problem
with flatbeds: not enough DR to "punch" through
a deep contrast image like Velvia produces.

I'm intensely curious as if these results will sustain
and keep with the 9k or if I'll be able to dig into the
dense spots more than with a 4990. I know the
raw scan will be a LOT sharper to start with - the
4990 has heaps of sharpness robbing glare.

This extra sharpness will make grain alias worse,
as 4000dpi is not enough to avoid it in the first
place. But I know how to get rid of that.
It's the DR that I'm deeply interested in comparing.

 

[Eddie]

I'd definitely recommend the Panasonic, I only upgraded to the E-500
when I found I couldn't get the control I wanted using a compact. I
still have the FZ7 though, it's great for video, and general shots. I'd
have a look at the outgoing FZ30 too, these are still a damn fine
camera.

Wow!Someone ELSE with an Olympus E500! With some of the criticism around I
thought I was the only one.

How do you find it (and please don't say by looking under the bed!) ;-)

Eddie

 

[Andy Hewitt]

[..]

Wow!Someone ELSE with an Olympus E500! With some of the criticism around I
thought I was the only one.

How do you find it (and please don't say by looking under the bed!) ;-)

Not at all, I have a dedicated cupboard ;-)

I did a lot of research first, and shortlisted the D50/D70, EOS350 and
of course the E500. I couldn't find a D50 to handle anywhere, and the
350 was just too small for my hands (although that had been my
favourite), the D50 I did see, didn't appear much bigger. I do like a
good sized camera - I used to fit the Winder 2 on my OM40 just for the
hand grip.

In the end though, the test shots I found revealed no great difference
in image performance for any of them, perhaps a slight edge for the
Canon - *if* you print images at full size, or crop a *lot*, or want
high ISO use. Of those I only use cropping, but not severely, and 8MP is
enough to deal with what I need.

In use, it's fantastic, at least compared to the compact. Still not as
much dynamic range as a film SLR, but they're stll good. I love the
lenses, and not having to worry about dust on the sensor is a great
bonus too.

I still have some of my old OM lenses, and I got an adapter off eBay for
a few pounds, which makes them still useful. Great thing is, I now have
a 75-250mm zoom lens, which has become a really useful telephoto on the
E500. I also kept the 50mm Zuiko F1.4. I did a test page of images from
each lens at different focal lengths, nobody could guess what was taken
with an old OM lens or the new 4/3 lenses.

I haven't really come across any down side as yet, although I do find
centre weighted metering is better than pattern metering, and Program is
better than Auto. Perhaps the only real issue I've found is a bit of
lens flare when shooting towards the sun, but what would do better?

 

[Raphael Bustin]

This:
http://members.iinet.net.au/~nsouto/photos/2006-10-02.jpg
is EXACTLY the same raw scanned file, but with the
grain aliasing removed. I did this by processing the raw
48-bit scanned file through NI and THEN vuescan it into
24-bit tiff and sharpened with the same settings as the other.
Ie, same workflow, same adjustments, only diff is I did filter
out the grain alias BEFORE I converted from 48-bit raw scan.

Sorry, if it's realy "grain aliasing," it can't be removed in post
processing, without also removing most of the underlying
detail. I thought you understood that. Anti-aliasing must
occur at the point of image capture.

I use and own NeatImage, but let's not kid ourselves.
This A:B comparison isn't proving what you think it does.

You're about to be disappointed, I fear. Even with
today's high resolutions (4000-5000 dpi) still interfere
with film grain to some extent. And most film scanners
(as far as I know) don't have AA filters.

Again, I'll be curious to see how the LS-9000 renders
this same hunk of film. I predict it will be deadly
sharp, and deadly noisy.

Same here. Grain aliasing is however something
that I've had to live with since scanning. It was
never there and I never saw it with enlargers.

No, not "aliasing" but the grain itself was sure
there, and the sharper the enlargement, the more
it showed. This was true even on 8x10" prints
made from Ektachrome and Kodachrome.

This is the demonstration I intended to show you
folks before. This time done within the exclusive
realm of scanned images so it's easier to compare.

I don't believe your "demonstration" proves what
you think it does, but no matter.

In order to show us scan that's truly *free* of
grain aliasing -- and yet still able to resolve the grain --
you'd need a scanner that has enough "excess"
resolution so that it can still resolve the grain even
through a proper AA filter. Or that its resolution
is so high that aliasing only begins at frequencies
well beyond those present in Velvia grain.



rafe b
www.terrapinphoto.com

 

[Eddie]

[..]

Wow!Someone ELSE with an Olympus E500! With some of the criticism around
I
thought I was the only one.

How do you find it (and please don't say by looking under the bed!) ;-)

Not at all, I have a dedicated cupboard ;-)

I did a lot of research first, and shortlisted the D50/D70, EOS350 and
of course the E500. I couldn't find a D50 to handle anywhere, and the
350 was just too small for my hands (although that had been my
favourite), the D50 I did see, didn't appear much bigger. I do like a
good sized camera - I used to fit the Winder 2 on my OM40 just for the
hand grip.

In the end though, the test shots I found revealed no great difference
in image performance for any of them, perhaps a slight edge for the
Canon - *if* you print images at full size, or crop a *lot*, or want
high ISO use. Of those I only use cropping, but not severely, and 8MP is
enough to deal with what I need.

In use, it's fantastic, at least compared to the compact. Still not as
much dynamic range as a film SLR, but they're stll good. I love the
lenses, and not having to worry about dust on the sensor is a great
bonus too.

I still have some of my old OM lenses, and I got an adapter off eBay for
a few pounds, which makes them still useful. Great thing is, I now have
a 75-250mm zoom lens, which has become a really useful telephoto on the
E500. I also kept the 50mm Zuiko F1.4. I did a test page of images from
each lens at different focal lengths, nobody could guess what was taken
with an old OM lens or the new 4/3 lenses.

I haven't really come across any down side as yet, although I do find
centre weighted metering is better than pattern metering, and Program is
better than Auto. Perhaps the only real issue I've found is a bit of
lens flare when shooting towards the sun, but what would do better?

Thanks, Andy. Got mine in Dubai a month or so back on way back to Oz.
Couldn't resit the two-lens package, though Canon are now doing something
similar here. Only thing I would like is the 14mm lens that they supply
here. Far more useful than the supplied 17.5 but then, canl't have
everything. I have a Canon G5 that is very good, indeed, excellent. But for
many pics shutter-lag kills it, not so with the Olympus, still a bit of lag
but small and doesn't stop me taking footy shots.

Eddie

 

[Andy Hewitt]

Thanks, Andy. Got mine in Dubai a month or so back on way back to Oz.
Couldn't resit the two-lens package, though Canon are now doing something
similar here. Only thing I would like is the 14mm lens that they supply
here. Far more useful than the supplied 17.5 but then, canl't have
everything. I have a Canon G5 that is very good, indeed, excellent. But for
many pics shutter-lag kills it, not so with the Olympus, still a bit of lag
but small and doesn't stop me taking footy shots.

I got the 14-45mm with mine here, a very good lens indeed. Some
retailers were offering the 17.5-45, but not making it very clear. It
makes a very good wide angle, although you do get a bit of vignetting
with filters attached.

It was shutter lag that I also found a problem with compacts, but the
E500 is almost as fast as my old OM40 - it actually manages the same
frames per second as the Winder 2.

I've always like Olympus stuff though, I started with an OM10, then got
an OM40, which I've only just sold on eBay, and my first digital was an
Olympus C960.

 

[Noons]

Sorry, if it's realy "grain aliasing," it can't be removed in post
processing, without also removing most of the underlying
detail. I thought you understood that. Anti-aliasing must
occur at the point of image capture.

Partly disagree. AA has been used in digital image
processing for many years now, at any stage of image
production without loss of anything. It used to be the norm in
the days when colour monitors were only 300X200X256: made
for some nice images instead of blotchy gunk. It still makes a
difference even in these days of much better monitors.

However, I agree entirely that when scanning, the sooner it is
introduced the better. Unfortunately as you well say: no modern
scanner hardware has that inbuilt. A pity, really. (or a market
opportunity for someone with the required technical ability).

NI makes a good job of filling in, provided it's introduced before
other processing takes place. I've tried using it at a later stage
and the results were never as satisfactory. But on the raw
vuescan output, it works rather well. I think they mention this
in their manuals/faq, can't recall where.

You're about to be disappointed, I fear. Even with
today's high resolutions (4000-5000 dpi) still interfere
with film grain to some extent. And most film scanners
(as far as I know) don't have AA filters.

Entirely agreed. No disappointment, I know
what I'm going to get in that area.

Again, I'll be curious to see how the LS-9000 renders
this same hunk of film. I predict it will be deadly
sharp, and deadly noisy.

Yup, it matches what I've seen so far. And I do
believe I now have a workflow solution for it. Dmax is still
something that worries me, though: I really hope the 9K
does a better job of it than flatbeds.

No, not "aliasing" but the grain itself was sure
there, and the sharper the enlargement, the more
it showed. This was true even on 8x10" prints
made from Ektachrome and Kodachrome.

Sure. Like I said: grain is not a problem.
Grain aliasing is.

In order to show us scan that's truly *free* of
grain aliasing -- and yet still able to resolve the grain --
you'd need a scanner that has enough "excess"
resolution so that it can still resolve the grain even
through a proper AA filter. Or that its resolution
is so high that aliasing only begins at frequencies
well beyond those present in Velvia grain.

A case in favour of the very high optical resolutions
of some drum scanners, IMO. I still think appropriate
AA filters at an early stage of the scanning process
would be a better - as in cheaper and more feasible - solution.

Wonder if the 7000dpi plustek gear would be enough to
get rid of aliasing? Must ask someone to run a few tests...
Their dmax is nothing to write home about but the rez
seems to be right up there.

 

[Raphael Bustin]

Wonder if the 7000dpi plustek gear would be enough to
get rid of aliasing? Must ask someone to run a few tests...
Their dmax is nothing to write home about but the rez
seems to be right up there.

I'm willing to wager that the Plustek claim is effectively
a doubled resolution (from using a staggered sensor) in
the same way the Epson flatbeds are doubled.

But that would mean (at least) that it's an oversampled
3600 dpi scanner which isn't necessarily a bad thing.

I really haven't seen a critical review of this beast but
I'm a bit suspicious because of its amazingly low price.


rafe b
www.terrapinphoto.com

 

[Raphael Bustin]

Partly disagree. AA has been used in digital image
processing for many years now, at any stage of image
production without loss of anything.

Completely disagree. It's a matter of ensuring the
Nyquist criteria. Without meeting that criteria, the
damage is done, and can't really be undone. Not
by Neat Image. Not by nuttin. No way, and no how.

You can do it anywhere *before* the sampling.

In this case, that it has to be implemented in the
optical path, in front of the CCD.

Anything else is post-processing.


rafe b
www.terrapnphoto.com