-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkl4OA4ACgkQlMjHkcbnFi2NcgCgkK0CPSzi1ZyamlaeV3IrFNyv
GNIAmgINgSmkCLJfEcIFKPK37tRFC8um
=zwlv
-----END PGP SIGNATURE-----

I have long preferred the Nikon scanners for scanning film, and your
post reinforces that.

20 minutes per scan? 1.0 to 1.5 GIGABYTE File sizes?

On an LS-2000 or LS-4000, it takes 2-4 minutes to scan (does not include
"futzing time" if you do manual cropping or exposure adjustments) , and
the file size is about 3 to 5 Megabytes for high quality JPEG or 50 MB
for TIFF. And I guarantee you that you won't find any quality
difference that you can see (not when you can see the film grain on the
Nikon scans ....)

I know that the Epsons are among the few flatbed scanners that CAN do a
good job (comparable in quality to the dedicated Nikon film scanners),
but I still maintain that overall, when you consider "efficiency" (which
includes a number of things) and ease of use, there is still no
comparison between those and the Nikon dedicated film scanners.

[I am assuming that we are talking about 35mm images; if you are
scanning larger images, then it's a moot point since the Nikon LS-2000
to 5000 scanner series can't scan larger images]


Gour wrote:
> Hi!
>
> I'm still experimenting which settings to use for my batch job...
>
> If one saves raw files in VueScan with the option 'Raw output with Scan'
> then the original CCD data are captured without any processing,
> i.e. "raw file has not had infrared-cleaning or grain reduction applied"
> which seems appropriate for archiving purposes.
>
> However, saving raw file as 64bit RGBi with infrared channel results in
> 2-pass scan which takes ~20mins for non-cropped 35mm slides on my
> laptop and results in 1.5G/1.0G files depending whether
> compression is used or not.
>
> I'm not so much concerned with the size (I'll use compression), but more
> with the scanning time which drops in case of saving as 48bit RGB to
> ~11mins (both times are for 4800dpi) which could be significant
> time-saver 8)
>
> Of course, I understand that in order to apply IR cleaning on my V700
> when doing post-processing work I need that extra IR channel, but
> wonder, whether the quality of Epson-V700 ICE and VueScan's
> infrared-cleaning feature justify investing in 'prolonged' scanning time
> and increased size of raw file, or put simply: should one save raw files
> as 48bit RGB or 64bit RGBi?
>
>
> Sincerely,
> Gour
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkl41TEACgkQlMjHkcbnFi0C7ACcD2zyQYk3aId5vAgbjvR9XjUs
0kQAn1hZ7jySiGiaCg7zWRWJRXbFzlyk
=ri1S
-----END PGP SIGNATURE-----

Going back to what I've said here MANY times .....

2,700 dpi scans of a 35mm slide produce a 10 megapixel image.

[The image size is 24mm x 36mm, which at 2,700 dpi is approximately
2,550 x 3,825 pixels, or 9.753750 megapixels; close enough to 10
megapixels for almost all purposes]

First, excluding a very small number of very professional images shot
with very high end equipment on very high quality film, there just isn't
any more than (or even) 10 megapixels worth of information present on a
typical consumer 35mm image shot on consumer film, with consumer
processing, with a consumer camera (even a good quality SLR) operated by
a non-professional photographer. Not to mention that in a lot of cases
the stuff we are scanning is 50 years old.

Second, if you do go above 2,700 dpi, total resolution (megapixels)
increases as the SQUARE of the resolution. So if you only go from 2,700
to 3,200 dpi, the resolution goes up to 14 megapixels, more than 40%
more, and that is just by going from 2,700 dpi to ONLY 3,200 dpi.

Consequently .... one can only really conclude that going to 4,800 or
6,400 dpi is an exercise in ignorant futility. 4,800 dpi scanning of a
35mm image generates over 30 megapixels of (mostly useless and
redundant) image data, and NO ONE would make an argument that ANY 35mm
film (or lens) can actually capture that much information), not
withstanding that current Epson and Nikon film scanners claim to be able
to scan at such resolutions.

[And, by the way, the optics (focus quality) of not only the camera that
shot the image but also of the scanner itself come into serious question
at such high resolutions. All considerations of the film and it's image
themselves not withstanding, and accepting that the Nikon and Epson
image sensors actually have 5,000 and 6,400 image sensor elements per
inch, respectively, that still begs the question: are the lenses in the
SCANNERS (either one) actually capable of resolving such fine detail,
even if it was present on the film (which, I maintain, it isn't)?

These firms (and many of their customers) are engaging in a war of
specifications that has become meaningless because they LONG AGO passed
the point at which any further increases in scanner resolution mattered.

Barry Watzman

Gour wrote:

>
> Well, let me do some more testing and compare 4800 vs 3200 dpi results.
>
> Thank you for your input.
>
>
> Sincerely,
> Gour
>
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkl935cACgkQlMjHkcbnFi1rBQCfSgDhf99ERkmVhQtmi3ln7cqw
9Y0AnjO3115WsNnElDdp+r/O/GOojmem
=STeE
-----END PGP SIGNATURE-----

There is another variable that we have not mentioned, which is color
depth. So you have {x} [mega] pixels. But in terms of file size, each
pixel can be anything from one bit (a pure black and white bitmap with
no color or even shades of gray) to at least 48 bits (6 bytes) per
pixel. It has a HUGE effect on the file size, obviously. In principle
an 8 megapixel image could range from 1 megabyte to 48 megabytes, quite
a range (like 48:1, Duh !!).

But what about the effect on perceived quality?

20 bits per pixel is exactly 1 million colors; 24 bits per pixel is
exactly 16 million colors. Experts differ on the exact number of colors
that the human eye can perceive (and presumably it varies from person to
person), but there is universal agreement that it is somewhere between 1
million and 10 million. So once you go beyond 24 bits per pixel, you
have clearly exceeded what is necessary to achieve reproduction of all
colors that humans can perceive. Yet, all of the latest scanners go WAY
beyond this ... well up into the 40+ bits per pixel.

I find it interesting that the "old" LS-2000 scanned at 2,700 dpi (10
megapixels) and also at 12 bits PER COLOR PER PIXEL (36 bits per pixel)
(it's poor stepsister cousin, the "dumbed down" LS-30, scanned at 10
bits per color (30 bits per pixel)). And yet Nikon (and Epson) keep
bumping these numbers up, with every generation: so current products
scan 35mm film images at 4,000 dpi and 48 bits per pixel.

Sorry, but I am not convinced that the hardware manufacturers are doing
us any favors. Enough is enough, and in fact, current products really
offer almost too much, especially in the hands of dumb, naive users who
will always use the maximum settings just because they are the maximum
settings and they THINK that they are getting some genuinely additional
quality when, in fact, all that they are doing is creating a 130+
megabyte FILE SIZE whose actual quality for any real-world purpose is no
better than that of a file one-quarter of that size.

[And there is a huge impact on scan time as well]


Gour wrote:

>
> I'm (and it seems I'm not alone) confused with stuff like: "he image
> quality is not good enough to compete against a good film scanner, the
> scanning speed is extremely slow, and the effective resolution of 2400
> dpi results in 7 megapixels for 35mm material, even though the files are
> extremely bloated with about 60 megapixels."[1]
>
> From which it looks like V700 gives 2400dpi quality, i.e. 7MP but one
> needs to scan at 6400dpi in order to get those bloated files in the
> range of 60MP?
>
> [1] See http://www.filmscanner.info/en/Epson...V700Photo.html review.
>
>
> Sincerely,
> Gour
>

On Sat, 24 Jan 2009 22:45:48 -0500, Barry Watzman
<(E-Mail Removed)> wrote:

-----------------clap trap removed

you are such a pompous self flogging arsehole. Get life.

#Barry Watzman

On Mon, 26 Jan 2009 23:33:07 -0500, Barry Watzman
<(E-Mail Removed)> wrote:

--------------------------------
#20 bits per pixel is exactly 1 million colors; 24 bits per pixel is
#exactly 16 million colors. Experts differ on the exact number of
colors
#that the human eye can perceive (and presumably it varies from person
to
#person), but there is universal agreement that it is somewhere
between 1
#million and 10 million.

last number published 2.4 mln (less for color blind).

nailer <(E-Mail Removed)> wrote:
> On Mon, 26 Jan 2009 23:33:07 -0500, Barry Watzman
> <(E-Mail Removed)> wrote:
>
> --------------------------------
> #20 bits per pixel is exactly 1 million colors; 24 bits per pixel is

Not "exactly 1 million colors", it's 2 to the power of 20 which is
1048576. It a fair approximation to 1000000 but hardly "exactly".

> #exactly 16 million colors. Experts differ on the exact number of
> colors
> #that the human eye can perceive (and presumably it varies from person
> to
> #person), but there is universal agreement that it is somewhere
> between 1
> #million and 10 million.
>
> last number published 2.4 mln (less for color blind).

--
Chris Green

Ok, guilty as charged.


(E-Mail Removed) wrote:
> nailer <(E-Mail Removed)> wrote:
>> On Mon, 26 Jan 2009 23:33:07 -0500, Barry Watzman
>> <(E-Mail Removed)> wrote:
>>
>> --------------------------------
>> #20 bits per pixel is exactly 1 million colors; 24 bits per pixel is
>
> Not "exactly 1 million colors", it's 2 to the power of 20 which is
> 1048576. It a fair approximation to 1000000 but hardly "exactly".
>
>> #exactly 16 million colors. Experts differ on the exact number of
>> colors
>> #that the human eye can perceive (and presumably it varies from person
>> to
>> #person), but there is universal agreement that it is somewhere
>> between 1
>> #million and 10 million.
>>
>> last number published 2.4 mln (less for color blind).
>