Digital ICE v FARE

[John]

Does Digital ICE work with positives as well as negatives or is it
just for negatives?

How much better are the Digital ICE scanners with optical density of
4.0D compared with the 3.4D ones? E.g. the 4490 compared to the 4990?

The Canon range of scanners that have the Fare/Qare(sp) IR film
correction/dust removal facility that is supposed to be similar to the
Digital ICE of the Epsons, I just wondered how it stacks up in
comparison?

When I have done a bit of research on the Canon websites I can't find
any specifications on what the optical densities are that the
different levels of Fare/Qare for the Canon Scanners.

I am presuming that seen as they don't mention this they are probably
nowhere near as good as the Epsons?

John

 

[Alan Browne]

Does Digital ICE work with positives as well as negatives or is it
just for negatives?

Both color negatives and positives, except kodachrome.

Ektachrome is fine.

Negative (really damaged), point at image to see scratch-correction.

Cheers,
Alan

How much better are the Digital ICE scanners with optical density of
4.0D compared with the 3.4D ones? E.g. the 4490 compared to the 4990?

Ick! Get a dedicated film scanner. Nikon CoolScan 5000, 9000 or a
used, non - II Minolta 5400; or even a Nikon V.

When I have done a bit of research on the Canon websites I can't find
any specifications on what the optical densities are that the
different levels of Fare/Qare for the Canon Scanners.

I am presuming that seen as they don't mention this they are probably
nowhere near as good as the Epsons?

If they're dedicated film, they're probably better than the Canon's.

For a quick idea of scan density: where n is the number of bits per
color channel.

Scan dens = log 2^(n-1.5). For 16 bits this = 4.36.

Cheers,
Alan

 

[Alan Browne]

Negative (really damaged), point at image to see scratch-correction.

Sorry, meant to include:

http://www.aliasimages.com/ScanScratch.htm

 

[tony sayer]

Sorry, meant to include:

http://www.aliasimages.com/ScanScratch.htm

Impressive

 

[Colin_D]

Does Digital ICE work with positives as well as negatives or is it
just for negatives?

How much better are the Digital ICE scanners with optical density of
4.0D compared with the 3.4D ones? E.g. the 4490 compared to the 4990?

Density has basically been hijacked as a sales term. Density is a
logarithmic function similar to decibels. Zero density is specified as
0.0D (clear film), a density of 4.0 implies a 10,000:1 density range
(10^4), while a density of 3.4 (10^3.4) implies a range of 2,511:1.
Given that Dmax of most films is about 2.0, both scanners have more than
adequate range. IMHO a D of 4.0 is advertising hype.

The Canon range of scanners that have the Fare/Qare(sp) IR film
correction/dust removal facility that is supposed to be similar to the
Digital ICE of the Epsons, I just wondered how it stacks up in
comparison?

FARE works as well, and is basically the same method as ICE in both
systems. Both work by scanning the film with IR light, to which film
dyes are transparent, so the resultant image contains IR-blocking items
like dust and scratches. This image is then used to fill in the
corresponding areas in the color image with color derived from adjacent
pixels.

When I have done a bit of research on the Canon websites I can't find
any specifications on what the optical densities are that the
different levels of Fare/Qare for the Canon Scanners.

I am presuming that seen as they don't mention this they are probably
nowhere near as good as the Epsons?

False presumption. I have a Canon 9950F, and it is every bit as good as
the Epsons. You can specify low, medium, or high settings for FARE. I
find that medium does all I require.

Colin D.

 

[Raphael Bustin]

Density has basically been hijacked as a sales term. Density is a
logarithmic function similar to decibels. Zero density is specified as
0.0D (clear film), a density of 4.0 implies a 10,000:1 density range
(10^4), while a density of 3.4 (10^3.4) implies a range of 2,511:1.
Given that Dmax of most films is about 2.0, both scanners have more than
adequate range. IMHO a D of 4.0 is advertising hype.

Umm, not quite. Slide films have Dmax values as high as 4 and
sometimes beyond. C41 (color print) film densities are lower,
usually peaking at 2.5 or so. Log scale, of course.

You can look it all up in the film data sheets; Fuji's are available
on the web.

So, with regard to scanners, C41 film is actually quite a bit
easier to scan, as it will almost never push a scanner beyond
its dynamic range. Slide film is another matter.


rafe b
www.terrapinphoto.com

 

[Barry Watzman]

I have seen examples that are even more impressive. This technology
really works, especially on color negatives.

 

[Toni Nikkanen]

FARE works as well, and is basically the same method as ICE in both
systems. Both work by scanning the film with IR light, to which film
dyes are transparent, so the resultant image contains IR-blocking
items like dust and scratches. This image is then used to fill in the
corresponding areas in the color image with color derived from
adjacent pixels.

I used to think that FARE was not as good as ICE, based on my experiences on a
dedicated 35mm film scanner with ICE vs. FARE on a cheapish Canon flatbed (8400F).
Later I have found the difference is probably in that the dedicated scanners just
work better; ICE on my Epson V700 is no miracle cure either, though it
works wonders in my Konica-Minolta 5400 II.

False presumption. I have a Canon 9950F, and it is every bit as good
as the Epsons. You can specify low, medium, or high settings for
FARE. I find that medium does all I require.

I found on my Canoscan 8400F that higher FARE settings cause general
softness in the image, instead of just around areas with dust or
scratches.

 

[Kennedy McEwen]

I used to think that FARE was not as good as ICE, based on my experiences on a
dedicated 35mm film scanner with ICE vs. FARE on a cheapish Canon
flatbed (8400F).
Later I have found the difference is probably in that the dedicated
scanners just
work better;

Even on dedicated film scanners, FARE is inferior to ICE.

The FS-4000US was Canon's only dedicated film scanner with FARE and it
differed significantly in its implementation from Nikon or Minolta ICE,
with resulting lower quality. One of the problems with the Canon is
that it used a separate pass of the scan head to create the IR image and
hence the film defects. Imperfections in the mechanism meant that there
was always misalignment between the IR and the visible images, which
required the cloned area around each defect to be much larger than with
ICE. FARE on Canon's dedicated film scanner often produced quite
visible corrections that ICE concealed perfectly on the Nikon units.

 

[John]

I used to think that FARE was not as good as ICE, based on my experiences on a

Even on dedicated film scanners, FARE is inferior to ICE.

The FS-4000US was Canon's only dedicated film scanner with FARE and it
differed significantly in its implementation from Nikon or Minolta ICE,
with resulting lower quality. One of the problems with the Canon is
that it used a separate pass of the scan head to create the IR image and
hence the film defects. Imperfections in the mechanism meant that there
was always misalignment between the IR and the visible images, which
required the cloned area around each defect to be much larger than with
ICE. FARE on Canon's dedicated film scanner often produced quite
visible corrections that ICE concealed perfectly on the Nikon units.

I think you're probably right about Fare v Digital ICE on dedicated
machines. Probably the implementation and the machines with ICE were
just better designed.

As far as flatbed scanners go I have been reading quite a few reviews
on different sites and so far I've arrived at the conclusion that
there isn't much difference between Fare and ICE as far as dust
removal is concerned. If anything though from this review I have read,
it seems that Fare may be slightly better just based on the fact that
it scans a lot faster using the dust removal facility than the Epson
machine with ICE.

http://www.photo-i.co.uk/Reviews/interactive/Scanners/Canon_9950F/page_14.htm

You'll have to scroll down to near the bottom of the page. The Epson
with ICE on took 7 and a half minutes whereas the Canon with Fare took
under 2 minutes. I think that would be the key selling point for me
despite the poor reputation of Canon here in the UK for support.

Maybe newer Epson models have improved with the speeds? This was just
comparing the Canoscan 9950F to the Perfection 4870. I think both
these models were from a few years ago, so maybe Epson speeds have
improved since then. Thing is though I don't have the money for a
V700/750 so it would still be out of the question anyway even if the
speeds had improved. I think I am going to be looking at a buying
used scanner, models like the Epson Perfection 4990, 4870 and 4490 as
well as the Canon Canoscans 9950F, 9900F, 8600F, 8400F etc. Probably
be able to afford the Canon model new.

A lot of the scans I have looked at the dedicated film scanners like
from the Nikon Coolscans the picture is a lot better I have to admit
that, even ones with lower resolution scans the picture is miles
better, so that proves to me that you shouldn't read too much into the
resolution specs. However I don't think you can scan regular prints
(positives) with these dedicated ones like the Coolscans can you? They
are just for negatives and slides right? For me I don't want to spend
big bucks on a dedicated scanner when the largest majority of the
scans I will be doing are from prints not negatives or slides.

Maybe I will just buy a flatbed for now for prints and you never know
I might get a dedicated film scanner for my old slides and negatives
at a later date if the flatbed proves not good enough.

John

 

[DenverDad]

Kennedy,

This is something I hadn't realized before - that the Nikons implement
the IR scan simultaneously with the visible scan, as opposed to
performing it separately. I am a Canon FS4000 user and had always
assumed that the difference between ICE/FARE was only in the
algorithms employed to fill-in the missing data. So this is very
interesting, as the potential for imperfect registration between scans
has always been very apparent with my FS4000. I have seen it's effect
when attempting to perform any kind of "long exposure" or multiple
pass technique with my scanner. And, I have struggled with many
versions of VueScan where strange IR scan artifacts are clear evidence
of a mismatch between the IR and visible scans. Simultaneous
scanning clearly has advantages!

Having said all that, my current settings on the FS4000 yield near-
perfect results as far as IR dust removal is concerned. With the
version of VueScan I use and the IR clean at the lowest setting, small
specks and defects are cleanly and consistently "removed" without any
loss of resolutiona. Now, since I don't have an ICE machine to
compare with, it is possible that I am still missing something and
could get better results with a Nikon 5000/9000 (and I certainly WOULD
get one if I could justify/afford it at this point!)

Anyway, do you know how the simultaneous visible/IR scan is
accomplished? Are there multiple sensors? Or is there perhaps just a
separate IR-filtered row on a single sensor array? And if so, does
this require a small "step" and a separate pass for each row? Or does
it somehow obtain visible and IR data truly simultaneously for each
pixel? Just courious!

Jeff

 

[Barry Watzman]

Simultaneous scanning has two advantages:

-elimination of registration errors
-It's faster (significantly faster)

I think that there is only one sensor, but that there are four different
light emitters (red, blue, green, IR). The light source in the Nikon
scanners is not a lamp but rather LEDs. So the sequence is something like:

-move the carriage
-red light source on, acquire scan data from CCD
-blue light source on, acquire scan data from CCD
-green light source on, acquire scan data from CCD
-IR light source on, acquire scan data from CCD
-move the carriage

 

[DenverDad]

Oh that's right, I forgot they use the LEDs instead of a lamp. Your
explanation makes sense. Well all I can say is that I wish my Canon
did it this way too. The time savings would be huge!

Simultaneous scanning has two advantages:

-elimination of registration errors
-It's faster (significantly faster)

I think that there is only one sensor, but that there are four different
light emitters (red, blue, green, IR). The light source in the Nikon
scanners is not a lamp but rather LEDs. So the sequence is something like:

-move the carriage
-red light source on, acquire scan data from CCD
-blue light source on, acquire scan data from CCD
-green light source on, acquire scan data from CCD
-IR light source on, acquire scan data from CCD
-move the carriage

Kennedy,

This is something I hadn't realized before - that the Nikons implement
the IR scan simultaneously with the visible scan, as opposed to
performing it separately. I am a Canon FS4000 user and had always
assumed that the difference between ICE/FARE was only in the
algorithms employed to fill-in the missing data. So this is very
interesting, as the potential for imperfect registration between scans
has always been very apparent with my FS4000. I have seen it's effect
when attempting to perform any kind of "long exposure" or multiple
pass technique with my scanner. And, I have struggled with many
versions of VueScan where strange IR scan artifacts are clear evidence
of a mismatch between the IR and visible scans. Simultaneous
scanning clearly has advantages!

Having said all that, my current settings on the FS4000 yield near-
perfect results as far as IR dust removal is concerned. With the
version of VueScan I use and the IR clean at the lowest setting, small
specks and defects are cleanly and consistently "removed" without any
loss of resolutiona. Now, since I don't have an ICE machine to
compare with, it is possible that I am still missing something and
could get better results with a Nikon 5000/9000 (and I certainly WOULD
get one if I could justify/afford it at this point!)

Anyway, do you know how the simultaneous visible/IR scan is
accomplished? Are there multiple sensors? Or is there perhaps just a
separate IR-filtered row on a single sensor array? And if so, does
this require a small "step" and a separate pass for each row? Or does
it somehow obtain visible and IR data truly simultaneously for each
pixel? Just courious!

 

[Kennedy McEwen]

Kennedy,

This is something I hadn't realized before - that the Nikons implement
the IR scan simultaneously with the visible scan, as opposed to
performing it separately. I am a Canon FS4000 user and had always
assumed that the difference between ICE/FARE was only in the
algorithms employed to fill-in the missing data. So this is very
interesting, as the potential for imperfect registration between scans
has always been very apparent with my FS4000. I have seen it's effect
when attempting to perform any kind of "long exposure" or multiple
pass technique with my scanner. And, I have struggled with many
versions of VueScan where strange IR scan artifacts are clear evidence
of a mismatch between the IR and visible scans. Simultaneous
scanning clearly has advantages!

Having said all that, my current settings on the FS4000 yield near-
perfect results as far as IR dust removal is concerned. With the
version of VueScan I use and the IR clean at the lowest setting, small
specks and defects are cleanly and consistently "removed" without any
loss of resolutiona. Now, since I don't have an ICE machine to
compare with, it is possible that I am still missing something and
could get better results with a Nikon 5000/9000 (and I certainly WOULD
get one if I could justify/afford it at this point!)

Anyway, do you know how the simultaneous visible/IR scan is
accomplished? Are there multiple sensors? Or is there perhaps just a
separate IR-filtered row on a single sensor array? And if so, does
this require a small "step" and a separate pass for each row? Or does
it somehow obtain visible and IR data truly simultaneously for each
pixel? Just courious!

Been away for a couple of days, hence the delay answering your question.

The simultaneous visible/IR capture, or more accurately single pass
visible/IR capture, is possible because of the unique method the the
Nikon scanners operate: rather than multiple sensors, the Nikon scanners
use a single line of sensors (except in their latest implementations
where additional lines of sensors are used to increase scan speed). Both
visible (R, G & B) and IR images are captured by pulsing different
colour LEDs to illuminate the film at each scan position.

It is the unique light source that makes single pass RGBI capture
possible with Nikon scanners, because LEDs can be switched on and off
extremely fast and reliably, whilst the cold cathode illumination of the
Canon scanner (and most others) can't, thus requiring a double pass to
capture the IR scan.

 

[Talker]

It is the unique light source that makes single pass RGBI capture
possible with Nikon scanners, because LEDs can be switched on and off
extremely fast and reliably, whilst the cold cathode illumination of the
Canon scanner (and most others) can't, thus requiring a double pass to
capture the IR scan.

Just curious but what does it cost to replace the scanning LEDs
in the Nikon scanner when they burn out?

Talker

 

[Barry Watzman]

The LEDs are not "scanning" LEDs they are just normal, stationary LEDs.
They have an expected life around 100,000 hours, and are lit only
during actual scanning. It would be very rare to ever have to replace them.

I see lots of advantages and no downsides to Nikons LED technology.

 

[Rob]

The LEDs are not "scanning" LEDs they are just normal, stationary LEDs.
They have an expected life around 100,000 hours, and are lit only
during actual scanning. It would be very rare to ever have to replace
them.

I see lots of advantages and no downsides to Nikons LED technology.

One downside is there not bright enough and the Nikon scanners don't
have as much DOF as some other scanners.

 

[Kennedy McEwen]

Just curious but what does it cost to replace the scanning LEDs
in the Nikon scanner when they burn out?

That is one of the advantages using LEDs instead of lamps - they don't
fail catastrophically, but slowly degrade to end of life which is orders
of magnitude higher than that of a lamp. The gradual degradation of
output is readily accommodated in the scanner by the auto-exposure
function. However with a rated life in the region of 100,000
operational hours to 50% output, the effect is marginal during the life
of the scanner.

Quite the opposite is true for lamps, which degrade during life but fail
dramatically at the end of life. In addition, whilst the lamp may have
a rated life of several tens of thousand hours, that is usually under
static conditions. Actual life is a lot less when the lamp is switched
on and off frequently during its life.

I have never heard of, and certainly have no experience of, a single
case of an LED in a Nikon scanner needing replacement. In contrast, I
have heard of hundreds of cases of scanner lamps needing replacement,
and have junked one scanner myself because the lamp burned out and it
was cheaper to buy a complete new scanner than to replace the lamp on
the old one.

 

[Toni Nikkanen]

Kennedy,
This is something I hadn't realized before - that the Nikons implement
the IR scan simultaneously with the visible scan, as opposed to
performing it separately.

It would seem the Epson V700 also needs to separate pass for the IR
channel; or that's at least how Silverfast AI implements it on that
scanner - I don't know how the Epson Scan software does it. Especially
on roll film and even larger formats this is not a good thing, because
the film is likely to change shape due to heat between scanning
passes.

 

[Raphael Bustin]

It would seem the Epson V700 also needs to separate pass for the IR
channel; or that's at least how Silverfast AI implements it on that
scanner - I don't know how the Epson Scan software does it. Especially
on roll film and even larger formats this is not a good thing, because
the film is likely to change shape due to heat between scanning
passes.

Don't know about the V700/750, but dICE on the Epson 4990
is done in a separate pass as well, using Epson's scan driver.
It caught me by surprise the first time I saw it in action.

rafe b
www.terrapinphoto.com