Is the Epson V700 / V750 the best compromise?

  • Thread starter Thread starter Tony
  • Start date Start date
A lot of the images I have looked at on an number of sites i-photo
included I have always found the ones the Coolscans have produced
better than Epson or Canon flatbeds including their newest and that
just backs up what people in this group have told me when I have asked
similar questions in the past.

Cheers

John
 
Kinon said:
For large format, what other choices do you have?
Click to expand...

I second that question, now that my Noblex 6x12 came back from repair.
How do I scan 6x12 cm. slides and negatives?

I think that a good flatbed scanner (e.g. Canon 9950F or Epson V700) is the
only available solution at the moment.

N.F.
 
I plan on getting a V750 at some point, but some reviewers say
it is not much different than an Epson 4990. The largest difference
I see in the better flatbeds versus dedicated film scanners is the
glass adds reflections and reduces contrast.
Click to expand...

No, the worst problem is the highly folded optical path
and the tiny, short-focal-length lenses used on flatbeds,
as compared to dedicated 35mm or MF scanners.

Compare the imaging lens on an LS-9000 to the one
on a 4990, or compare the optical paths on the two
-- it's like comparing Palomar to a Questar.

That layer of glass that your negative sits on is really
only a minor part of the problem.
However, scanning
at 16-bits/pixel it is easily corrected.
Click to expand...

I've worked enough with CCDs and scanner firmware to
hazard a guess that anything beyond 11-12 bits of real
S/N from a CCD is a pipe dream.


rafe b
www.terrapinphoto.com
 
I plan on getting a V750 at some point, but some reviewers say
it is not much different than an Epson 4990. The largest difference
I see in the better flatbeds versus dedicated film scanners is the
glass adds reflections and reduces contrast.
Click to expand...

No, the worst problem is the highly folded optical path
and the tiny, short-focal-length lenses used on flatbeds,
as compared to dedicated 35mm or MF scanners.

Compare the imaging lens on an LS-9000 to the one
on a 4990, or compare the optical paths on the two
-- it's like comparing Palomar to a Questar.

That layer of glass that your negative sits on is really
only a minor part of the problem.
However, scanning
at 16-bits/pixel it is easily corrected.
Click to expand...

I've worked enough with CCDs and scanner firmware to
hazard a guess that anything beyond 11-12 bits of real
S/N from a CCD is a pipe dream.


rafe b
www.terrapinphoto.com
 
Raphael said:
No, the worst problem is the highly folded optical path
and the tiny, short-focal-length lenses used on flatbeds,
as compared to dedicated 35mm or MF scanners.

Compare the imaging lens on an LS-9000 to the one
on a 4990, or compare the optical paths on the two
-- it's like comparing Palomar to a Questar.
Click to expand...

Hmmm 200 inch/3.5 inch = 57 times the aperture.
I have taken apart some flatbeds and in the better ones,
lens aperture is ~3/4 inch or so. I haven't checked
a 4990, of dedicated film scanner. You're saying
a dedicated scanner has lenses much larger than 3/4 inch
(0.75 * 57 = 42-inch aperture lens ;-)?
That layer of glass that your negative sits on is really
only a minor part of the problem.
Click to expand...

If the aperture, f/ratio and sensor deliver the MTF, then
the optical system is not the issue. There are quite large
linear arrays available at low cost these days, so matching
the linear sensor to a lens system is not that hard.
I've worked enough with CCDs and scanner firmware to
hazard a guess that anything beyond 11-12 bits of real
S/N from a CCD is a pipe dream.
Click to expand...

To get 12-bits of S/N one needs to collect over 16 million
photons (with zero read noise), so I'll agree with you
here. With read noise of 4 electrons, you would need to
collect over 65 million photons/pixel! No CCD or CMOS
sensor can do that, unless the pixel is mm in diameter.
Are you referring to dynamic range?

Then with image stacking (multiple sampling in scanner),
one can beat down the noise and get well above 12-bits dynamic range.
This is done even with 8-bit web cams by amateur astronomers
imaging the planets! Multisampling scanners are no big deal.
Film doesn't have 12-bits of S/N, not even 200 (perhaps if you
average large areas with special controlled development).
I think you mean dynamic range.

Roger
 
Hmmm 200 inch/3.5 inch = 57 times the aperture.
I have taken apart some flatbeds and in the better ones,
lens aperture is ~3/4 inch or so. I haven't checked
a 4990, of dedicated film scanner. You're saying
a dedicated scanner has lenses much larger than 3/4 inch
(0.75 * 57 = 42-inch aperture lens ;-)?
Click to expand...

I was exaggerating a bit. But seriously, the lenses are
hugely different. I'm guessing most flatbeds use something
like a 20 mm focal length, f/8 lens. And about four
mirror bounces in the optical path. The Nikon
scanner uses a 75-80 mm 14-element ED lens with
what looks like an f/3.5 aperture. And exactly one
mirror bounce. The Epson lens assembly weighs about
5 grams, the Nikon is probably about 200 grams.

I have requested V750 scan samples on numerous
occasions for the "scan snippets" page. Nobody's
come forward.

Even better would be a direct V750 vs LS-8000
comparison, though that would take a tiny bit more
effort. Suffice to say the Nikon is still comfortably
ahead of the Epson V700.

I think you mean dynamic range.
Click to expand...

I think I mean bits that represent real information
rather than noise. I'm saying that the limited S/N
of CCD (and means of capturing CCD data) means
that the upper 6 or 7 bits (in a 16 bit capture)
are mostly noise.


rafe b
www.terrapinphoto.com
 
Raphael said:
I was exaggerating a bit. But seriously, the lenses are
hugely different. I'm guessing most flatbeds use something
like a 20 mm focal length, f/8 lens. And about four
mirror bounces in the optical path. The Nikon
scanner uses a 75-80 mm 14-element ED lens with
what looks like an f/3.5 aperture.
Click to expand...

Hmmm... I bet the f/8 lens has better performance ;-).
The point is that a 14-element lens working at f/3.5
may not necessarily perform better on a 2d slide than
a simpler lens at f/8 with a larger linear sensor.

And exactly one
mirror bounce. The Epson lens assembly weighs about
5 grams, the Nikon is probably about 200 grams.
Click to expand...

The mirrors will not affect image quality in any meaningful
way.
I have requested V750 scan samples on numerous
occasions for the "scan snippets" page. Nobody's
come forward.
Click to expand...

Yes, it would be nice to see.
Even better would be a direct V750 vs LS-8000
comparison, though that would take a tiny bit more
effort. Suffice to say the Nikon is still comfortably
ahead of the Epson V700.
Click to expand...

I don't dispute this. I would like to see examples too.
I think I mean bits that represent real information
rather than noise. I'm saying that the limited S/N
of CCD (and means of capturing CCD data) means
that the upper 6 or 7 bits (in a 16 bit capture)
are mostly noise.
Click to expand...

Signals anywhere above the lowest couple of stops are
photon noise limited. And that is directly proportional
the the pixel size, with larger pixels collecting more photons.
What size CCD is in a Nikon LS-8000?

Remember flatbeds are designed to do large film too.
E.g., doesn't the V750 do 6000 ppi over 4x5 inches.
That's 6000*4 = 24,000 pixels wide! Of which,
I would guess 10,000 pixels gives ~50% MTF.

Roger
 
In comp.periphs.scanners Raphael Bustin said:
I have requested V750 scan samples on numerous
occasions for the "scan snippets" page. Nobody's
come forward.

Even better would be a direct V750 vs LS-8000
comparison, though that would take a tiny bit more
effort. Suffice to say the Nikon is still comfortably
ahead of the Epson V700.
Click to expand...
So no one has done a direct comparison but you still say "the Nikon is
still comfortably ahead of the Epson V700.". :-)
 
Ole-Hjalmar Kristensen said:
Look at the the water in the upper part. The difference is pretty
obvious there. The differences in the boat are perhaps less visble.
Click to expand...

Also, if the Nikon picture is unsharpened as the file naming would
imply, it could be much improved further by unsharp masking, just as
the Epson picture has been.
 
HiYa


it is not much different than an Epson 4990. The largest difference
I see in the better flatbeds versus dedicated film scanners is the
glass adds reflections and reduces contrast. However, scanning
Click to expand...

hmm ... I seem to recall reading a www page showing the differences on this,
just recently tried googling for it with no luck.

do you happen to have the link handy? (don't bother ferreting if you don't)

I've been thinkin of doing some home brew modifications to my 4870 and
spending the bucks on a LS-4000 2nd hand for 35mm ... just to get the best of
both worlds.

I've been wondering some major hacking and about making up something
comparable to a flextight curved holder ... but I've been pondering the
ability to have machined parts made up by a mate with the right gear.

problem is that now I've got the time, the scanner is in a different part of
the world to me!

dratt!

See Ya
(when bandwidth gets better ;-)

Chris Eastwood
Photographer, Programmer
Motorcyclist and dingbat

please remove undies for reply
 
Terve Toni

I'd add two other points:
- Flatbeds don't have an autofocus mechanism (not that I know of; height-
adjustable film holders do exist, though)
- Flatbeds don't seem to read RGB and IR channels in the same pass; this
contributes to the factor that Digital ICE doesn't work as well as in
a dedicated film scanner
Click to expand...

not to mention the different light source (I don't think they're using the
cathode to generate the IR)

certainly my 4870 ICE is hardly worth writing home about.
Also, flatbeds have 2 more glass surfaces to keep clean. Of course,
using the glass carrier of the Nikon or Minolta Multipro means you
have four extra glass surfaces to keep clean!
Click to expand...

wonder if the dust makes itself visible at those distances (out of the focal
range).

on another note, where should I be looking around for second hand scanning
gear here in Finland? Huuto has no automated notifications like eBay, and eBay
leaves me dealing with Germans and Pomms. The Germans seem to want to get
revenge for being ousted in WW2 with the postage rates, and the Pomms seem
reluctant to use anything except Royal Mail (which costs too much too)


See Ya
(when bandwidth gets better ;-)

Chris Eastwood
Photographer, Programmer
Motorcyclist and dingbat

please remove undies for reply
 
As the fine print says, it's the 4870...
Click to expand...

yeah ... moments after I posted, I scrolled down ... voila!

too late to withdraw the post :-)

I don't know how come I've missed this page before. If I see differences like
that on scans of 35mm negative (and why not?) with a Nikon 4000 then I might
defer buying anything more in the DSLR range and stick with just my 10D and my
film body. Unless you know where I can get a 5D for "go-man-en" ;-)


See Ya
(when bandwidth gets better ;-)

Chris Eastwood
Photographer, Programmer
Motorcyclist and dingbat

please remove undies for reply
 
Obakesan said:
I don't know how come I've missed this page before. If I see differences
like
that on scans of 35mm negative (and why not?) with a Nikon 4000 then I
might
defer buying anything more in the DSLR range and stick with just my 10D
and my
film body. Unless you know where I can get a 5D for "go-man-en" ;-)
Click to expand...

My take on the Nikon 8000 (the same technology as the 4000) is that by noise
reducing and downsampling (and sharpening) I can create files that are
roughly as clean and sharp and noise free as 5D files. (I should put up some
examples...)

The only question is: how far do I have to downsample?

I find that I can reliably get very nice files by NR and downsampling to
2200 ppi. From a 6x7 frame, this is 27 very tasty MP. More than twice the
5D. Killer 16x20 prints. But from 35mm it's barely 6MP. (I should put up
some examples...)

IMHO, it's conceiveably that one could do somewhat better.

But the 5D is essentially a 3000 ppi scan of 24x36mm at 5D quality.

I simply do not believe that one is going to get 3000 ppi 5D quality pixels
from scans of film.

David J. Littleboy
Tokyo, Japan
 
HiYa

"David J. said:
My take on the Nikon 8000 (the same technology as the 4000) is that by noise
reducing and downsampling (and sharpening) I can create files that are
roughly as clean and sharp and noise free as 5D files. (I should put up some
examples...)
Click to expand...

really? which format? Surely not 35mm??

I find that I can reliably get very nice files by NR and downsampling to
2200 ppi. From a 6x7 frame, this is 27 very tasty MP. More than twice the
5D.
Click to expand...

ohh ... probably you mean from that one...
But from 35mm it's barely 6MP. (I should put up
some examples...)
Click to expand...

well, if you have time, yes please :-) but don't break into schedules for it
I simply do not believe that one is going to get 3000 ppi 5D quality pixels
from scans of film.
Click to expand...


me neither (I suspect that the 4000dpi scans downsampled to 2700 and compared
to the 5D might come close to a downsampled 5D image :-) But hey, I often eat
mashed potatoes and mince balls made at home too cos its cheaper

When I can justify spending the bucks on a 5D (or equivalent) I'll do it, but
(as you know) I juggle a few formats and that one ( [35mm|Digital] SLR) just
doesn't get used as much as my compact digital and my 4x5. Besides the
argument that a 1D might get as good a result as my 4x5 (for me) the money
spend isn't worth the gains.

anyway, we're getting into 'personal bias' territory, and that's all different
equations to maths

thanks for the update


See Ya
(when bandwidth gets better ;-)

Chris Eastwood
Photographer, Programmer
Motorcyclist and dingbat

please remove undies for reply
 
Back
Top