More on scanning 126 negative strips and slides.

[John Corliss]

I contacted Epson and asked if they could recommend anything for this.
Their response was that they don't "have any scanner models that
specifically support 126 film which is 4.25" x 6.5"."

Not sure what they meant by that, because a 126 negative film strip
isn't 4 1/4" tall. I had been very clear in my request that I was
talking about *negatives*, not *prints*.

They went on to say that the closest they have under $1000 is the
Perfection 4870 model which can scan 4" x 5" negatives but would
require that the negatives be cut. Cutting up family owned 126
negative film stips isn't an option for me.

Also not an option for me are scanners over say, $250.

I don't understand why all the scanner manufacturers think that the
only size film that everybody has ever used is 35 mm. Hell, when I was
a kid, 35mm film was a luxury. Even color film was initially only for
those with bucks (I'm 53 right now.) Then 126 came along and was VERY
popular; Kodak made a lot of money off of that format. My folks used
it almost exclusively and I would really like to be able to digitize
those images.

Is there some problem I'm not aware of that makes it impossible for
scanner manufacturers to make an affordable slide scanner with just a
*slightly* larger scan area (Christ, we're only talking about a few
square millimeters here!) so that 126 negative strips and slides can
also be done?

The Epson 4870 currently costs from $370 on up. I can't afford that.
I've also checked out the Polaroid SprintScan 35 and from what I can
see, I might have to cut up the strips to scan the images, and this is
simply not an option. Also, there are two versions of the scanner, the
"LE":

http://www.cix.co.uk/~tsphoto/tech/filmscan/pol35le/pol35le.htm

and the "Plus":

http://www.cix.co.uk/~tsphoto/tech/filmscan/pol35plus/pol35plu.htm

Obviously, the Plus is the preferred version because of its much
higher resolution. However, I don't know what an "optional pathscan
enabler" is. Maybe this would make doing what I want possiblem
(assuming I could also buy one from eBay or someplace else.)

At any rate, what is the problem with finding a simple way to scan 126
negatives and slides? It was a VERY popular film format from the late
60s until sometime in the 80s. Does anybody know why this should be so
difficult? Any further suggestions?

TIA

 

[CSM1]

I contacted Epson and asked if they could recommend anything for this.
Their response was that they don't "have any scanner models that
specifically support 126 film which is 4.25" x 6.5"."

Not sure what they meant by that, because a 126 negative film strip isn't
4 1/4" tall. I had been very clear in my request that I was talking about
*negatives*, not *prints*.

They went on to say that the closest they have under $1000 is the
Perfection 4870 model which can scan 4" x 5" negatives but would require
that the negatives be cut. Cutting up family owned 126 negative film stips
isn't an option for me.

Also not an option for me are scanners over say, $250.

I don't understand why all the scanner manufacturers think that the only
size film that everybody has ever used is 35 mm. Hell, when I was a kid,
35mm film was a luxury. Even color film was initially only for those with
bucks (I'm 53 right now.) Then 126 came along and was VERY popular; Kodak
made a lot of money off of that format. My folks used it almost
exclusively and I would really like to be able to digitize those images.

Is there some problem I'm not aware of that makes it impossible for
scanner manufacturers to make an affordable slide scanner with just a
*slightly* larger scan area (Christ, we're only talking about a few square
millimeters here!) so that 126 negative strips and slides can also be
done?

The Epson 4870 currently costs from $370 on up. I can't afford that.
I've also checked out the Polaroid SprintScan 35 and from what I can see,
I might have to cut up the strips to scan the images, and this is simply
not an option. Also, there are two versions of the scanner, the "LE":

http://www.cix.co.uk/~tsphoto/tech/filmscan/pol35le/pol35le.htm

and the "Plus":

http://www.cix.co.uk/~tsphoto/tech/filmscan/pol35plus/pol35plu.htm

Obviously, the Plus is the preferred version because of its much higher
resolution. However, I don't know what an "optional pathscan enabler" is.
Maybe this would make doing what I want possiblem (assuming I could also
buy one from eBay or someplace else.)

At any rate, what is the problem with finding a simple way to scan 126
negatives and slides? It was a VERY popular film format from the late 60s
until sometime in the 80s. Does anybody know why this should be so
difficult? Any further suggestions?

TIA

It is difficult because the 126 film format has been dead for many, many
years.

If you want to spend the money, get a Medium format film scanner, then make
a mask to fit the 126 negatives.

A Flatbed Scanner with 2 1/4 inch transparency adapter will work. You still
have to make your own mask for the 126 film. Not hard to do with cardboard.


An Epson Perfection 3170 PHOTO is about $200 and will do 2 1/4 inch
transparencies.

 

[RSD99]

First: "John Corliss" posted:
"...Not sure what they meant by that, because a 126 negative film strip
isn't 4 1/4" tall. I had been very clear in my request that I was
talking about *negatives*, not *prints*.
...."

There is some confusion on the web page "Classic Camera Film Sizes,
Sources, and Film Adapters" by Robert Monaghan
http://medfmt.8k.com/bronfilms.html#126
where he gives two different dimensions for "Kodak Instamatic 126-format
cameras" sized film. His first table (which is apparently in error) shows a
film size of "4 1/4 x 6 1/2 inches" ... which I think is actually the print
size. I think the "Technical Support" people at Epson probably got their
information from this source. However, the second table shows a more
realistic film size of 28 mm x 28 mm, which would be roughly 1.102" x
1.102". This would be a little small for the Epson 4870, but in reality
should scan without any problems. It just will not fit into any of the
Epson supplied "film holders."

Just to make sure we are all talking about the same film size, could you
measure the image area of one of your negatives, and post this information
back to this thread.

Second: "John Corliss" posted:
"...
They [Epson] went on to say that the closest they have under $1000 is the
Perfection 4870 model which can scan 4" x 5" negatives but would
require that the negatives be cut. Cutting up family owned 126
negative film stips isn't an option for me.
...."

That's not really true. The Epson 4870 would probably scan the 126
negatives quite well. However, it might not provide the resolution you
want, as the effective resolution actually is more like 1000 to 1600
samples per inch.

You would only have to cut the negatives if you felt that it was necessary
to use Epson's supplied "film holders." Instead, just place the negatives
emulsion side down on the scanner's glass, cover with a piece of
Anti-Newton-Ring glass (to flatten them), and scan away.

You could probably also scan these negatives using some of the lower cost
Epson flatbed scanners, such as the new 4180 or the 3170. The film scanners
you mentioned will certainly give better results, but it would probably
require something like a glass carrier in one of the medium format scanners
such as the Nikon Coolpix 9000 to cover the 28 mm square format.

FWIW:
Epson's scanner page is
http://www.epson.com/cgi-bin/Store/ProductCategory.jsp?BV_UseBVCookie=yes&o
id=-8172

 

[Stephen Harker]

I contacted Epson and asked if they could recommend anything for
this. Their response was that they don't "have any scanner models that
specifically support 126 film which is 4.25" x 6.5"."

Not sure what they meant by that, because a 126 negative film strip
isn't 4 1/4" tall. I had been very clear in my request that I was
talking about *negatives*, not *prints*.

As has been mentioned by other posters Instamatic 126 film negatives
are 28x28mm. The film strips are the same as 35mm film. I scanned my
old 126 Instamatic negatives in my 35mm film scanner. Since 35mm film
is 36x24mm the film scanners normally just use a scan strip width of
24mm and scan lengthways. When used for 126 Instamatic film this
meant I missed a strip of 4mm at the top. I had to move the film
strip after every two scans to put avoid the vertical bar negative
separator on the film strip holder between two negatives.

Mostly this was acceptable for me. You should be able to scan the
full negative using any flat bed scanner with a 35mm film adapter.

I don't understand why all the scanner manufacturers think that the
only size film that everybody has ever used is 35 mm. Hell, when I was
a kid, 35mm film was a luxury. Even color film was initially only for
those with bucks (I'm 53 right now.) Then 126 came along and was VERY
popular; Kodak made a lot of money off of that format. My folks used
it almost exclusively and I would really like to be able to digitize
those images.

I think it is the usual question of market size trade-offs. For 35mm
film scanners it is cheaper to have a small sensor and possibly
technically easier to design the sensors in that way. The market for
these machines is relatively high end. There are film scanners that
can do more formats, but they are even more expensive. Most of the
flat bed scanners that can do negatives will do 126 Instamatic, many
will do medium format as well, though at a higher cost point.

 

[John Corliss]

First: "John Corliss" posted:
"...Not sure what they meant by that, because a 126 negative film strip
isn't 4 1/4" tall. I had been very clear in my request that I was
talking about *negatives*, not *prints*.
..."

There is some confusion on the web page "Classic Camera Film Sizes,
Sources, and Film Adapters" by Robert Monaghan
http://medfmt.8k.com/bronfilms.html#126
where he gives two different dimensions for "Kodak Instamatic 126-format
cameras" sized film. His first table (which is apparently in error) shows a
film size of "4 1/4 x 6 1/2 inches" ... which I think is actually the print
size. I think the "Technical Support" people at Epson probably got their
information from this source. However, the second table shows a more
realistic film size of 28 mm x 28 mm, which would be roughly 1.102" x
1.102". This would be a little small for the Epson 4870, but in reality
should scan without any problems. It just will not fit into any of the
Epson supplied "film holders."

Just to make sure we are all talking about the same film size, could you
measure the image area of one of your negatives, and post this information
back to this thread.

For the negatives, the total film strip height is the same as 35 mm
(35 mm) but the actual image size is 28 mm x 28 mm.

The *printed* image size (obviously one could purchase enlargements,
but this was the "standard size" print) was either 8 cm x 8 cm (with a
white border not added in) or, if borderless was 8.6 cm 8.8 cm. In
inches, that's 3.15 inches x 3.15 inches for the first, 3.39 inches x
3.46 inches for the second.

I haven't a clue what Robert Monaghan was referring to with his 4 1/4"
to 6 1/2" measurements. That must be a reference to the original 126
type dating back to 1906. What I'm talking about is 126 instamatic
format. Obviously the guy at Epson got it wrong.

Second: "John Corliss" posted:
"...
They [Epson] went on to say that the closest they have under $1000 is the
Perfection 4870 model which can scan 4" x 5" negatives but would
require that the negatives be cut. Cutting up family owned 126
negative film stips isn't an option for me.
..."

That's not really true. The Epson 4870 would probably scan the 126
negatives quite well. However, it might not provide the resolution you
want, as the effective resolution actually is more like 1000 to 1600
samples per inch.

*Sigh* Sounds like yet another example of deceitful advertising. I
remember when there was a class action lawsuit against monitor
companies for advertising the tube size instead of the actual viewing
size. They got into hot water for that. Now it's scanners and
"interpolated resolution" vs optical or true resolution. Just curious
though, where did you get that effective resolution rating for the 4870?

You would only have to cut the negatives if you felt that it was necessary
to use Epson's supplied "film holders." Instead, just place the negatives
emulsion side down on the scanner's glass, cover with a piece of
Anti-Newton-Ring glass (to flatten them), and scan away.

I found this site:

http://home.earthlink.net/~dougfisher/holder/mainintro.html

but I'm sure there are others. Thanks for the tip.

You could probably also scan these negatives using some of the lower cost
Epson flatbed scanners, such as the new 4180 or the 3170. The film scanners

I just got another email from Epson and they recommend that model:

"The 3170 is specified as being able to scan 35mm film, unfortunately
other sizes are not supported. The Perfection 4180 costs the same as
the 3170 after the current mail-in rebate and will scan this size as
well as medium format if desired."

you mentioned will certainly give better results, but it would probably
require something like a glass carrier in one of the medium format scanners
such as the Nikon Coolpix 9000 to cover the 28 mm square format.

FWIW:
Epson's scanner page is
http://www.epson.com/cgi-bin/Store/ProductCategory.jsp?BV_UseBVCookie=yes&o
id=-8172

Thanks, but I already have it bookmarked. And thanks for responding!

 

[John Corliss]

It is difficult because the 126 film format has been dead for many, many
years.

Well, at least since 1999. And when you're my age, that seems like
yesterday. Also, there are lots and lots of people with hundreds of
126 negatives that they'd like to digitize.

If you want to spend the money, get a Medium format film scanner, then make
a mask to fit the 126 negatives.

Ah, but I *don't* want to spend the money. In fact, I can't since I
don't have it.

A Flatbed Scanner with 2 1/4 inch transparency adapter will work. You still
have to make your own mask for the 126 film. Not hard to do with cardboard.

I agree and am looking into that possiblity. A friend of mine has a
3170, but Epson claims (in an email I got yesterday) that it won't do
126 negatives. I'm going over to my friend's house and give it a try.

An Epson Perfection 3170 PHOTO is about $200 and will do 2 1/4 inch
transparencies.

Thanks for replying.

 

[John Corliss]

As has been mentioned by other posters Instamatic 126 film negatives
are 28x28mm. The film strips are the same as 35mm film. I scanned my
old 126 Instamatic negatives in my 35mm film scanner. Since 35mm film
is 36x24mm the film scanners normally just use a scan strip width of
24mm and scan lengthways. When used for 126 Instamatic film this
meant I missed a strip of 4mm at the top. I had to move the film
strip after every two scans to put avoid the vertical bar negative
separator on the film strip holder between two negatives.

Not sure what you mean by this last part.

Mostly this was acceptable for me.

That's the problem. It's about the same for me, but in many cases, the
image loss is unacceptable and I don't want to hassle with stitching
two scans together.

You should be able to scan the
full negative using any flat bed scanner with a 35mm film adapter.

Correct me if I'm wrong, but don't 35 mm film adapters cut off the
image the same way as film scanners do?

I think it is the usual question of market size trade-offs. For 35mm
film scanners it is cheaper to have a small sensor and possibly
technically easier to design the sensors in that way. The market for
these machines is relatively high end. There are film scanners that
can do more formats, but they are even more expensive. Most of the
flat bed scanners that can do negatives will do 126 Instamatic, many
will do medium format as well, though at a higher cost point.

I'm also looking at the Epson 4180 Photo right now. According to a
second email I just got from Epson, it can do the job.

Thanks very much for your input.

 

[CSM1]

Not sure what you mean by this last part.


That's the problem. It's about the same for me, but in many cases, the
image loss is unacceptable and I don't want to hassle with stitching two
scans together.


Correct me if I'm wrong, but don't 35 mm film adapters cut off the image
the same way as film scanners do?


I'm also looking at the Epson 4180 Photo right now. According to a second
email I just got from Epson, it can do the job.

Thanks very much for your input.

Correct me if I'm wrong, but don't 35 mm film adapters cut off the image
the same way as film scanners do?

Film Scanners put the film strip into a holder that has cutouts for the
image area of the 35mm film, a 24mm X 36mm cutout.

35 mm format has sprocket holes on the top and bottom of the film strip.

126 format is 28 mm X 28mm square and the sprocket holes are on the bottom
only. The top of the image goes almost to the top of the film.

A comparison of the two types of film strips. FYI these were scanned at a
low res on a flatbed scanner with a 5" X 7" transparency adapter built into
the lid. They are scanned as transparencies, not negatives. I wanted to
include the orange mask and not reverse the image.
http://www.carlmcmillan.com/Temp/

"As always, Mr. Phelps, the tape will be destroyed in 5 seconds."
I mean temporary. As in not going to be available for long.

On flatbeds, it depends on how the Adapter is constructed. If the Adapter is
just a light that covers the 35 mm film width, then it would work fine for
scanning 126 format.

The Epson 3170 Photo is a good choice for this job because the light source
is built into the lid and is about 2.6 inches wide.
You would use the Pro mode to manually set the image area to the square
frame of the 126 film.

 

[RSD99]

asked:
"...
Just curious
though, where did you get that effective resolution rating for the 4870?
...."

Several places ... including

(1) The scanner bake-off
http://www.jamesphotography.ca/scanner_test.html

(2) My own Epson Perfection 4870

First: "John Corliss" posted:
"...Not sure what they meant by that, because a 126 negative film strip
isn't 4 1/4" tall. I had been very clear in my request that I was
talking about *negatives*, not *prints*.
..."

There is some confusion on the web page "Classic Camera Film Sizes,
Sources, and Film Adapters" by Robert Monaghan
http://medfmt.8k.com/bronfilms.html#126
where he gives two different dimensions for "Kodak Instamatic 126-format
cameras" sized film. His first table (which is apparently in error) shows a
film size of "4 1/4 x 6 1/2 inches" ... which I think is actually the print
size. I think the "Technical Support" people at Epson probably got their
information from this source. However, the second table shows a more
realistic film size of 28 mm x 28 mm, which would be roughly 1.102" x
1.102". This would be a little small for the Epson 4870, but in reality
should scan without any problems. It just will not fit into any of the
Epson supplied "film holders."

Just to make sure we are all talking about the same film size, could you
measure the image area of one of your negatives, and post this information
back to this thread.

For the negatives, the total film strip height is the same as 35 mm
(35 mm) but the actual image size is 28 mm x 28 mm.

The *printed* image size (obviously one could purchase enlargements,
but this was the "standard size" print) was either 8 cm x 8 cm (with a
white border not added in) or, if borderless was 8.6 cm 8.8 cm. In
inches, that's 3.15 inches x 3.15 inches for the first, 3.39 inches x
3.46 inches for the second.

I haven't a clue what Robert Monaghan was referring to with his 4 1/4"
to 6 1/2" measurements. That must be a reference to the original 126
type dating back to 1906. What I'm talking about is 126 instamatic
format. Obviously the guy at Epson got it wrong.

Second: "John Corliss" posted:
"...
They [Epson] went on to say that the closest they have under $1000 is the
Perfection 4870 model which can scan 4" x 5" negatives but would
require that the negatives be cut. Cutting up family owned 126
negative film stips isn't an option for me.
..."

That's not really true. The Epson 4870 would probably scan the 126
negatives quite well. However, it might not provide the resolution you
want, as the effective resolution actually is more like 1000 to 1600
samples per inch.

*Sigh* Sounds like yet another example of deceitful advertising. I
remember when there was a class action lawsuit against monitor
companies for advertising the tube size instead of the actual viewing
size. They got into hot water for that. Now it's scanners and
"interpolated resolution" vs optical or true resolution. Just curious
though, where did you get that effective resolution rating for the 4870?

You would only have to cut the negatives if you felt that it was necessary
to use Epson's supplied "film holders." Instead, just place the negatives
emulsion side down on the scanner's glass, cover with a piece of
Anti-Newton-Ring glass (to flatten them), and scan away.

I found this site:

http://home.earthlink.net/~dougfisher/holder/mainintro.html

but I'm sure there are others. Thanks for the tip.

You could probably also scan these negatives using some of the lower cost
Epson flatbed scanners, such as the new 4180 or the 3170. The film

scanners

I just got another email from Epson and they recommend that model:

"The 3170 is specified as being able to scan 35mm film, unfortunately
other sizes are not supported. The Perfection 4180 costs the same as
the 3170 after the current mail-in rebate and will scan this size as
well as medium format if desired."

you mentioned will certainly give better results, but it would probably
require something like a glass carrier in one of the medium format scanners
such as the Nikon Coolpix 9000 to cover the 28 mm square format.

FWIW:
Epson's scanner page is
http://www.epson.com/cgi-bin/Store/ProductCategory.jsp?BV_UseBVCookie=yes&o
id=-8172

Thanks, but I already have it bookmarked. And thanks for responding!

 

[John Corliss]

Film Scanners put the film strip into a holder that has cutouts for the
image area of the 35mm film, a 24mm X 36mm cutout.

35 mm format has sprocket holes on the top and bottom of the film strip.

126 format is 28 mm X 28mm square and the sprocket holes are on the bottom
only. The top of the image goes almost to the top of the film.

A comparison of the two types of film strips. FYI these were scanned at a
low res on a flatbed scanner with a 5" X 7" transparency adapter built into
the lid. They are scanned as transparencies, not negatives. I wanted to
include the orange mask and not reverse the image.
http://www.carlmcmillan.com/Temp/

"As always, Mr. Phelps, the tape will be destroyed in 5 seconds."
I mean temporary. As in not going to be available for long.

Thank you very much for the clarification. Nice job on the page!

On flatbeds, it depends on how the Adapter is constructed. If the Adapter is
just a light that covers the 35 mm film width, then it would work fine for
scanning 126 format.

The Epson 3170 Photo is a good choice for this job because the light source
is built into the lid and is about 2.6 inches wide.
You would use the Pro mode to manually set the image area to the square
frame of the 126 film.

Many thanks again. This helps a lot.

 

[John Corliss]

asked:
"...
Just curious
though, where did you get that effective resolution rating for the 4870?
..."

Several places ... including

(1) The scanner bake-off
http://www.jamesphotography.ca/scanner_test.html

Many thanks. That will be a great reference to have. Bookmarked.

(2) My own Epson Perfection 4870

So then, are you saying that the advertised resolution is only the
interpolated type or does post scanning processing actually decrease
the optical resolution to create the image? Epson claims that the
resolution is 4800 x 9600 dpi but don't state whether that is true
optical or not (at least that I could find.)

 

[Fernando]

*Sigh* Sounds like yet another example of deceitful advertising. I
remember when there was a class action lawsuit against monitor
companies for advertising the tube size instead of the actual viewing
size. They got into hot water for that. Now it's scanners and
"interpolated resolution" vs optical or true resolution. Just curious
though, where did you get that effective resolution rating for the 4870?

In this case, it's not an "interpolated" resolution false advertising,
like it was common in the past years.
The Epson 4870 CCD would actually be able to read at 4800dpi (well,
someone would say that due to its peculiar sensor architecture it
would not reach "real" 4800dpi, but this is not the point) , but the
optical system (glass bed + mirror + fixed-focus small-aperture lens)
fails to give it enough details to scan.
Epson can throws any hyper-hyped sci-fi CCD sensor she likes, but if
the optical system remains the same, it would not go beyond actual
1600-2000 "real" dpi, unfortunately. And besides, it suffers from
limited dynamic range. Not a show-stopper for negative scans, but for
slides it's a shame.

Fernando

 

[Kennedy McEwen]

Many thanks. That will be a great reference to have. Bookmarked.


So then, are you saying that the advertised resolution is only the
interpolated type or does post scanning processing actually decrease
the optical resolution to create the image? Epson claims that the
resolution is 4800 x 9600 dpi but don't state whether that is true
optical or not (at least that I could find.)

No, he isn't saying it is interpolated at all - although, to be fair, he
isn't actually saying anything particularly clearly or being
particularly accurate and fair either.

The issue is one of definitions - what resolution actually means. The
OED has around 10 different definitions, and at least two of these are
relevant to scanners - both of which are confused into a single sentence
in RSD99's statement.

Resolution(1): the highest spatial frequency that an imaging sensor can
resolve (resolve: distinguish between two separate objects and a single
extended one), measured in cycles per unit length (mm, inches, feet,
miles etc.) or angle.

Resolution(2): the minimum step that a mechanism can make, measured in
steps or samples per unit length.

You need at least two of these resolution(2) samples to support each
resolution(1) cycle and, whilst having more doesn't hurt, having less
causes immense problems.

Scanner manufacturers always quote resolution of the second type, and
are legally obliged to state if this is "interpolated" resolution.
Interpolated resolution means that the mechanism itself cannot make
steps as small as those quoted and the electronics are guessing what the
signal would be if it could.

The problem comes when they use the term "optical resolution". Most
people would consider this to be the first type of resolution - a
measure of what the scanner can optically resolve. But it isn't, and
you can blame some dimwit lawyers who agreed this definition some time
ago to distinguish it from "interpolated" resolution. So scanner
manufacturers use the term when what they really mean is "mechanical
resolution" - this is the physical separation of the individual real
samples that the scanner makes. Since these are real samples that the
scanner makes at every pixel position and they are not interpolated or
guessed, this is not an interpolated resolution. In fact, the Epson
4870 can produce interpolated resolution up to 9600ppi - and even then
it is only interpolating in one axis, two out of every four samples are
real! In terms of signal to noise ratio on the image, these real
samples add real information which can be measured, interpolated samples
do not add any additional information at all.

However, just as you can have an excellent resolution lens on a camera
and produce very low resolution results if you can't focus it or hold it
steady, so having a scanner with a high mechanical resolution does not
mean that the resolution (of the first type) is guaranteed. The scanner
might have great mechanics and lousy optics. Legally, however, the
scanner manufacturer is permitted to claim that the "optical resolution"
is the sampling density or mechanical resolution because it is not
interpolated. That is the legal distinction - even if it is highly
misleading.

These days, almost all flatbed scanners have a true optical resolution
(of the first type) which is significantly lower than what the
mechanical resolution can support. This is not necessarily a bad thing,
since it is important that the true optical resolution is less than the
Nyquist of the mechanical resolution, or the sampling density. Which is
where RDS99's comments go astray - not only quoting resolution(1) in
resolution(2) units, but in overestimating what the desired resolution
actually is or should be.

The quoted "optical resolution" (ie. mechanical resolution) of the Epson
4870 is 4800ppi. Mike Engles plot of the performance of his 4870
scanner at
http://www.jamesphotography.ca/bakeoff2004/Epson_4870_4800_auto_mike_engl
es_MTF.jpg (watch the text wrap) shows an MTF falling to zero at around
0.35cy/pixel, as opposed to the maximum desirable limit of 0.5cy/pixel.
Consequently the true resolution is equivalent to around 0.35/0.5 x
4800ppi, or a 3300ppi scanner - almost twice the resolution that RDS99
claims. These claims seem to be based on the MTF20 figure, which at
0.1432cy/pix equates to the sampling limit of a 1375ppi scanner.
However if you look at the MTF20 figure for the best of the scanners in
the 2004 "bake-off", it is 0.1997cy/pix for the Minolta5400, a 5400ppi
"optical resolution" device. Therefore, using the same metric as RDS99,
this "best of breed" device would only rate as a 2157ppi device. Nobody
in their right mind would consider that to be an accurate representation
of the Minolta 5400, so why should anyone consider it to be an accurate
representation of the Epson? It just isn't.

In other words, compare apples with apples and oranges with oranges. In
comparison to what everyone considers are real 4000ppi and 5400ppi film
scanners, the Epson 4870 actually achieves something equivalent to a
3000-3300ppi scanner. This is still less than its perfectly legally
claimed optical resolution might lead you to expect but considerably
better than the figures that RDS99 has given you when compared to what
the best scanners on the market actually achieve.

Don't just consider numbers in isolation - and remember there is a lot
more to scanner performance than just resolution. The dynamic range of
most flatbeds is inferior to a true film scanner, even though the Dmax
figure may be similar - now that *is* deceptive advertising, but they
all seem to quote the equivalent digital limit, whether their analogue
circuits can match it or not.

 

[RSD99]

"John Corliss" posted:
"...
So then, are you saying that the advertised resolution is only the
interpolated type or does post scanning processing actually decrease
the optical resolution to create the image?
...."

That's actually the resolution of the sensor. You can actually get 4800
unique pixels per inch out of the machine ... but of those 4800 pixels, it
takes something like 6 to 9 pixels to go from black to white on a knife
edge test. Hence, the actual end result is that the "Resolution" ... in
terms of what you can see on the target image ... is somewhat less than the
stated 4800 "dpi."This is not a factor of the "post scanning process," but
more a result of the optical limits of the flatbed scanner configuration,
with a few other things like focus error(s), reflections, flare, and so
forth thrown in for good measure.

Unfortunately, it's a subject beyond my ability to accurately describe to
you. However, the subject has been discussed several times (by others) here
in this forum. You might try looking through the Google Groups archives if
you want to pursue it further.

"John Corliss" then posted:
"...
Epson claims that the resolution is 4800 x 9600 dpi but don't state whether
that is true optical or not (at least that I could find.)
...."

That's the "mechanical resolution" ... to "coin a phrase." The 4870 is
actually a fine scanner, possibly one of the best (if not currently *the*
best) consumer level scanners available, and the somewhat misleading
resolution statements by Epson are typical of the limitations of the
specifications of *all* flatbed scanners, by *all* manufacturers. See above
for more information.

FWIW:
The horizontal resolution of 4800 dpi is based on the number of sensors in
the scanner, and the 9600 dpi for the vertical dpi is based on how many
steps the scanner's stepper motor can make.

 

[RSD99]

"Kennedy McEwen" posted:
"...
<Snip>
Mike Engles plot of the performance of his 4870
scanner at
http://www.jamesphotography.ca/bakeoff2004/Epson_4870_4800_auto_mike_engl
es_MTF.jpg (watch the text wrap) shows an MTF falling to zero at around
0.35cy/pixel, as opposed to the maximum desirable limit of 0.5cy/pixel.
Consequently the true resolution is equivalent to around 0.35/0.5 x
4800ppi, or a 3300ppi scanner - almost twice the resolution that RDS99
claims.
...."

I fervently wish *my* Epson Perfection 4870 could produce a scan of a knife
edge that was that clean.

Kennedy ... I really think that you need to consider that *some* Epson 4870
scanners perform better than others. Apparently, mine is in the latter
category ... but then it's been repaired by a field service (independent)
facility, and may not actually be performing to "factory specifications."

 

[David J. Littleboy]

The quoted "optical resolution" (ie. mechanical resolution) of the Epson
4870 is 4800ppi. Mike Engles plot of the performance of his 4870
scanner at
http://www.jamesphotography.ca/bakeoff2004/Epson_4870_4800_auto_mike_engl
es_MTF.jpg (watch the text wrap) shows an MTF falling to zero at around
0.35cy/pixel, as opposed to the maximum desirable limit of 0.5cy/pixel.
Consequently the true resolution is equivalent to around 0.35/0.5 x
4800ppi, or a 3300ppi scanner - almost twice the resolution that RDS99
claims. These claims seem to be based on the MTF20 figure, which at
0.1432cy/pix equates to the sampling limit of a 1375ppi scanner.
However if you look at the MTF20 figure for the best of the scanners in
the 2004 "bake-off", it is 0.1997cy/pix for the Minolta5400, a 5400ppi
"optical resolution" device. Therefore, using the same metric as RDS99,
this "best of breed" device would only rate as a 2157ppi device. Nobody
in their right mind would consider that to be an accurate representation
of the Minolta 5400, so why should anyone consider it to be an accurate
representation of the Epson? It just isn't.

In other words, compare apples with apples and oranges with oranges. In
comparison to what everyone considers are real 4000ppi and 5400ppi film
scanners, the Epson 4870 actually achieves something equivalent to a
3000-3300ppi scanner. This is still less than its perfectly legally
claimed optical resolution might lead you to expect but considerably
better than the figures that RDS99 has given you when compared to what
the best scanners on the market actually achieve.

I think you are still overestimating the 4870. Here's a page that
compares the Japanese version of the 4870 with a 2700 dpi scanner. The 4870
is clearly _worse_ than the 2700 (or is it 2800???) dpi scanner.

http://www5e.biglobe.ne.jp/~longnose/scanner_test.html

I suspect that careful postprocessing of 4870 images could
result in decent quality 2400 dpi scans that would support good 8x
enlargements.

For medium format, this would be pretty reasonable: good A3 and decent
quality 13x19
prints from 645.

Don't just consider numbers in isolation - and remember there is a lot
more to scanner performance than just resolution. The dynamic range of
most flatbeds is inferior to a true film scanner, even though the Dmax
figure may be similar - now that *is* deceptive advertising, but they
all seem to quote the equivalent digital limit, whether their analogue
circuits can match it or not.

Yes, although I suspect the problem is more in the optics than in the
circuitry. I found that the 2450 was a lot less obnoxious for negatives than
for slides (and the 6x7 and 6x9 folks were quite happy with the 2450), so
perhaps sticking to negative materials and medium format is required for
being happy with the 4870.

David J. Littleboy
Tokyo, Japan

 

[Kennedy McEwen]

Kennedy ... I really think that you need to consider that *some* Epson 4870
scanners perform better than others. Apparently, mine is in the latter
category ... but then it's been repaired by a field service (independent)
facility, and may not actually be performing to "factory specifications."

I think *you* need to consider the implications of that last sentence
before using your "experience" to comment on relative performance of
certain devices to others.

Presumably your service was conducted with a warranty - you have the
evidence that it is not performing as it should, so take it back and
tell them to do the job properly!

 

[Kennedy McEwen]

I think you are still overestimating the 4870. Here's a page that
compares the Japanese version of the 4870 with a 2700 dpi scanner. The 4870
is clearly _worse_ than the 2700 (or is it 2800???) dpi scanner.

The LS-40 is actually 2900ppi as stated in the article itself, however
it is interesting to note that on the automatic setting, the Epson
produces a scan that is much better than Andreas achieved manually - and
only marginally inferior to the Nikon.

I suspect that careful postprocessing of 4870 images could
result in decent quality

Remember that the whole principle of oversampled detectors is that the
results must be post scan sharpened specifically to recover the reduced
mid-range frequencies reproduced by the larger CCD elements with lower
noise. The concept of comparing an unprocessed output with that of a
linear array is simply miss-placed logic.

 

[RSD99]

"Kennedy McEwen" posted:
"...
I think *you* need to consider the implications of that last sentence
before using your "experience" to comment on relative performance of
certain devices to others.
...."

Then I think you really *need* to review the results of the Scanner
Bake-Off
http://www.jamesphotography.ca/bakeoff2004/scanner_test_results.html
Where the results were

Owner MTF 10% - 90%

Mike Engles 15.4 5.66 pixels ** The "sample"
you picked **

Mike Schlanker 12.69 6.98 pixels

Jeff Cable 11.5 7.52 pixels

Doug Fisher 11.1 8.31 pixels

Chris Thomas 8.995 9.97 pixels

Just a reminder ... my statement was

"... as the effective resolution actually is more like 1000 to 1600 samples
per inch. ..."

Which I believe is born out by the above data. Unfortunately, not all Epson
Perfection 4870 owners are achieving the absolute best performance level,
as experienced by Mike Engles [the example you erroneously used]... some of
us are seeing performance more like that experienced by "Chris Thomas" ...
and the "average" performance of an Epson 4870 would probably be pretty
close to the midpoint of the test data (MTF = 12.2, 10% - 90% 7.8 pixels).

Note that there's roughly a 75% spread in the data.

 

[Kennedy McEwen]

Unfortunately, not all Epson Perfection 4870 owners are achieving the
absolute best performance level, as experienced by Mike Engles [the
example you erroneously used]...

No, it was not erroneously picked. It was deliberately picked as the
best result for that scanner produced and compared with the best result
for the best of the dedicated film scanners. If you want to make
comparative statements then you have to use the same baseline to do so
and I made no secret of the fact I was using the best results from both
scanners to assess the relative performance.

Note that there's roughly a 75% spread in the data.

Whoop-de-doo! It might surprise you that there is a spread in all
scanners - and in scanner operators! That is why I used the best
results from both - to compare like for like, whilst noting that your
"assessment" did not even consider what a real practical scanner could
achieve by comparison, just some wild assumption of what you think an
ideal scanner should do!

Your statement is not borne out by the data at all on any like for like
basis. Even using the median figures of all of the scanner results for
each type of scanner produces a result for the Epson as being equivalent
to almost double the range you quoted it in.