Using Epson 4990 to scan 6X9 cm negs

[Bill]

Hello
I'm working with a friend who shoots a Fuji 6X9 wide and has made
terrific C prints up to 20X30 with negatives of Agfa Optima. He's
beginning to scan the negs with an Epson 4990 and, from the start,
they are very "grainy" (cloudy) and, even at 4800 dpi, don't hold up
well even at 11X14". Last night we ran some tests: 4800 dpi, no
filtration in the scanner (grain reduction, scratch removal, ICE etc)
and target size, original. The results, when sized to, say, 300 dpi
are quite "grainy" and fully hideous at 100% view. We'd like to
produce a 15X30" print on an Epson 4800. Our sense is that these scans
are barely 15% the quality of his C prints. I know that people here
would suggest a better scanner but, shouldn't there be some fairly
useable quality from a 4990?
We're really looking for some parameters for these scans.
Thanks,
Bill

 

[Ken Weitzel]

Hello
I'm working with a friend who shoots a Fuji 6X9 wide and has made
terrific C prints up to 20X30 with negatives of Agfa Optima. He's
beginning to scan the negs with an Epson 4990 and, from the start,
they are very "grainy" (cloudy) and, even at 4800 dpi, don't hold up
well even at 11X14". Last night we ran some tests: 4800 dpi, no
filtration in the scanner (grain reduction, scratch removal, ICE etc)
and target size, original. The results, when sized to, say, 300 dpi
are quite "grainy" and fully hideous at 100% view. We'd like to
produce a 15X30" print on an Epson 4800. Our sense is that these scans
are barely 15% the quality of his C prints. I know that people here
would suggest a better scanner but, shouldn't there be some fairly
useable quality from a 4990?
We're really looking for some parameters for these scans.

Hi Bill...

Don't have a 4990; still stuck back at a 3200 photo, but all I can
suggest is that something's terrible terrible wrong.

No idea what - how 'bout putting a sample scan (or a crop of one) up
somewhere so we can have a look see?

Take care.

Ken

 

[Bob AZ]

Hello
I'm working with a friend who shoots a Fuji 6X9 wide and has made
terrific C prints up to 20X30 with negatives of Agfa Optima. He's
beginning to scan the negs with an Epson 4990 and, from the start,

Send me a neg and I will try it on my Nikon 4500 scanner and print it
on my 4800 if it scans well.

Bob AZ

 

[thomas.c.monego]

Hello
I'm working with a friend who shoots a Fuji 6X9 wide and has made
terrific C prints up to 20X30 with negatives of Agfa Optima. He's
beginning to scan the negs with an Epson 4990 and, from the start,
they are very "grainy" (cloudy) and, even at 4800 dpi, don't hold up
well even at 11X14". Last night we ran some tests: 4800 dpi, no
filtration in the scanner (grain reduction, scratch removal, ICE etc)
and target size, original. The results, when sized to, say, 300 dpi
are quite "grainy" and fully hideous at 100% view. We'd like to
produce a 15X30" print on an Epson 4800. Our sense is that these scans
are barely 15% the quality of his C prints. I know that people here
would suggest a better scanner but, shouldn't there be some fairly
useable quality from a 4990?
We're really looking for some parameters for these scans.
Thanks,
Bill

Couple of things to try,
1) Make sure the negs are emulsion side up, illogical but it is how
the three Epson flatbeds I have used work.
2) Try a little lower res at 16 bit say 2400 or 3200, turn off
sharpening. By going to 4800 you maybe accentuating the grain. Do your
sharpening carefully in Photoshop.
3) Like above, on my V700 I don't see much difference between 4800 and
3200 ppi, I'm suspecting the res maxes out at 3200ppi on the V700.
The scans I am doing on my V700 are prtinting on a Canon iPF5000
better than any CPrint I have had done, comparible to CibaChromes I
have had printed by a Ciba specialist. Have to admit I don't shoot
color negs very often.

Tom

 

[Bill]

Hi Ken,
Thanks for your interest. I will try to post a crop this afternoon.
The original scan is 350MB or so. If I can make a good representation
as a lower rex jpeg that still shows the noise, would that work?
Bill

 

[Bill]

Hi Bob,
Thanks so much for your offer. We may take you up on that a little
later.
Bill

 

[Bill]

Hi Ken,
Here's a link to a scan sample. I posted many particulars below the
image.
Any help appreciated,
Bill

http://www.stephenlehmer.com/ScanTests/4800t.htm

 

[Colin_D]

Hello
I'm working with a friend who shoots a Fuji 6X9 wide and has made
terrific C prints up to 20X30 with negatives of Agfa Optima. He's
beginning to scan the negs with an Epson 4990 and, from the start,
they are very "grainy" (cloudy) and, even at 4800 dpi, don't hold up
well even at 11X14". Last night we ran some tests: 4800 dpi, no
filtration in the scanner (grain reduction, scratch removal, ICE etc)
and target size, original. The results, when sized to, say, 300 dpi
are quite "grainy" and fully hideous at 100% view. We'd like to
produce a 15X30" print on an Epson 4800. Our sense is that these scans
are barely 15% the quality of his C prints. I know that people here
would suggest a better scanner but, shouldn't there be some fairly
useable quality from a 4990?
We're really looking for some parameters for these scans.
Thanks,
Bill

I think your problem, in a word, is aliasing. Grain aliasing, probably
because a flat-bed scanner, even though it can scan film, has no
anti-aliasing filter. This is a longish read, but worth-while if you
want to know more:

http://www.photoscientia.co.uk/Grain.htm

Colin D.

 

[1Scan]

I'd like to express my appreciation to Bill for the original question
and to Colin D for pointing to such an informative answer.

I too have an Epson 4990 and the image Bill posted is consistent with
scans I've produced in similar circumstances. The technical answer
seems to me to be that the scanner is doing too good a job, in that
its resolving the details of the film and not the universally smooth
colour of the sky. The article suggests the use of Kodak's GEM
function as a solution. I have done a quick scan on my Nikon Coolscan
5000 from a 35mm slide and it seems this could be a solution.

I haven't tested it but using Silverfast scanning software its
possible to introduce a degree of enlargement in the scanning process.
Maybe selecting a lower dpi figure combined with a level of
magnification would overcome the problem. However I've found using
this function in Silverfast massively increases scan times.

Silverfast Ai also has a function called GANE - Grain and Noise
Elimination. Comparing the sample on Bill's site with the images in
Taz Tally's manual on Silverfast they look the same, so that too could
be a way of overcoming this issue. The GANE tool includes a useful
before and after view window so you should be able to see quickly if
this is solving the problem. I believe you can download a trial
version of Silverfast from their website.

Many thanks for the pointer to such an important issue.

Jeff Underwood
1Scan.co.uk

 

[Bill]

Thanks for everyone's responses. This has been one of those nice
University of the Net experiences. Tonight, we're gonna try lower
resolution, around, 1700 and some Neat Image for the first time.
Testing looks good so far. Does anyone have any suggestions for when
to sharpen with NI? ..before noise reduction? after?
Thanks all,
Bill

 

[-]

Does anyone have any suggestions for when

to sharpen with NI? ..before noise reduction? after?

You can do some sharpening within Neat Image (third options tab) during its
processing but I usually do Neat Image as one of the first steps in my image
processing workflow. Final/major unsharp masking is usually just about the
last step in my image workflow.

My $.02,
Doug

 

[Ole-Hjalmar Kristensen]

The colors seem strange. Also the contrast seems low. Have you tried
the auto exposure function? The noise you see is grain. I see similar
grain im my own scans with a 4990, but I also see it in the negative
with a loupe, so it is not my scanner who is making it up, althoug I
have see examples of grain aliasing. If you want to get rid of the
grain, try something like NeatImage.

 

[Ole-Hjalmar Kristensen]

From the 4990 I would not expect significantly more detail by scanning
at more than 2400 dpi, but sampling at 4800 and downsampling yourself
to 2400 *may* give better results than scanning at 2400 if the scanner
uses an inferior downsampling algorithm or simply does not use the
full resolution of its sensor array while scanning at lower resolution.