Archived from groups: alt.comp.periphs.scanner,rec.photo.film+labs (
More info?)
In article <_NqdnR-1ypHHfEDdRVn-gw@speedfactory.net>, Michael A.
Covington <look@ai.uga.edu.for.address> writes
>
>"Roger Halstead" <Delete-Invallid.groups@tm.net> wrote in message
>news:268rd0lanrbocm577d3ai83fc46iltnd50@4ax.com...
>
>> >My Coolscan III captures everything that's on the slide. In fact I
>usually
>> >use it at half resolution (1350 dpi) to hide grain. Remember that
>digital
>>
>> 2700 dpi should not be capable of capturing any where near all the
>> information on the slide unless it's very high speed film.
>
>2700 dpi is more than 100 lines per mm, more than 50 line pairs per mm.
>Photographs taken under normal conditions seldom resolve this much.
>
MTF charts for Velvia show an MTF of 50% at 50cy/mm, for Provia & Sensia
it is more typically 30% (as it is for the E-200 you refer to below),
which is certainly not negligible and a long way short of their limiting
resolution.
>Under
>ideal conditions, an excellent lens will resolve 80 line pairs per mm, but
>that's exceptional.
A decent lens will easily have an MTF of around 80% on axis for 50cy/mm
at f/8 or faster. At the frame edges and corners it would typically be
down on this, but not to the point of being negligible.
So, a tripod mounted camera shooting Fuji Velvia should easily reproduce
detail on the film at a transfer function of around 40% of the scene at
the limiting resolution of your LS-30 scanner. With a 2700ppi scanner,
the MTF at 50cy/mm is unlikely to exceed 50%, so 50% of the information
recorded on the film at the scanner's limiting resolution is lost - even
more if you halve the intrinsic resolution of the scan to 1350ppi! In
either case, that still leaves a significant amount of detail recorded
on the film that the scanner cannot reproduce without aliasing.
>
>> How things look at 2700 or 1350 depends more on the physical
>> characteristics of the scanner, but you should not have to scan at
>> half resolution to hide grain unless it's coming from the scanner and
>> not the slide.
>
>It's real film grain (Ektachrome 200, typically).
>
If its E-200 and an LS-30 then it definitely is *not* real grain! Just
have a true chemical 16x20" print made from any of your slides and
compare that to a similar size inkjet, or digital/chemical print created
from a scan off your LS-30 and the difference in grain size, structure
and perception between the two images is striking.
I certainly am not talking theoretically here Michael, I have actually
done this comparison both with slides and negatives, having made the
prints in my own darkroom with traditional methods.
The vast majority of the grain you see on the 2700ppi scan is aliased by
the scanner itself - and the light system and MTF curves of the Nikon
scanner range make them particularly prone to this. Its visibility
will, of course, be reduced by correctly downsampling to 1350ppi as your
workflow implements, but not without significant loss of real image
content as well.
Moving to the 4000ppi of the LS-4000 from my immediate predecessor
scanner the LS-2000 (same resolution as the LS-30, just more bits and
better features) the single biggest change in scan quality was the
perception of grain due, in the most part, to a massive reduction in
grain aliasing. Increased resolution was less significant but very
noticeable nonetheless. You can print the 4000ppi image directly
without need to downsample and get a perceived granularity very similar
to the chemical print. By applying a little filtering, such as GEM or
NeatImage (not too much!) you can get a print from the scan which is
actually cleaner than any chemical print yet with almost as much
resolution. Furthermore, 4000ppi is not even the limit of what can be
on the film with a good lens setup, as users of the Minolta 5400 scanner
will testify and can be verified with a professional drum scan. Such
results are simply impossible from 2700ppi scanners, good though some of
them indisputably were.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a ah heck when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)