Several weeks ago Gary Reese offered one of his test slides to the list
for two bucks. I ordered one of these slides and Gary suggested that I
report my results back to the list. Perhaps others will be inspired to
do the same. (Come on, Tom, we need to know how that 4000 dpi Xerox
actually performs).
I case you're not familiar with Gary's test setup: The test shot is a
US Geological Survey map of the Grand Canyon. National Bureau of
Standards resolution charts are taped to the center and corners of the
map. The resolution charts consist of multiple, progressivly smaller
and smaller targets with both horiziontal and vertical line patterns.
The targets are marked with numbers which must then be multiplied by the
magnification factor (41) to arrive at the actual resolution on the film
in line pairs per millimeter (lppm). The charts (in the range relevant
to this discussion) are labeled 1.0, 1.1, 1.25, 1.4, 1.6, 1.8, 2.0,
2.2. This particular test slide is taken with a Zuiko 50mm f/1.4 at
f/5.6 and shows both center AND corner resolution of 90 lppm. This is
determined by finding the smallest target where one can count the lines
(even though they may not be 100 leanly resolved along their entire
length). With suitable microscope magnification this is found to be the
chart labeled 2.2. Multiplying by 41 yields 90 lppm.
The first thing I did after receiving the slide was to attempt to verify
the 90 lppm resolution of the test slide. I dragged out an old student
microscope I had bought in a garage sale about 15 years ago which had
been hiding in the closet ever since. I first reported to Gary that the
scope was only able to show me 1.8 (74 lppm) but I was using too strict
a standard expecting the lines to be 100% resolved along their entire
length. I also realized that the scope was dreadfully dirty and the
illuminator didn't work since the battery box was full of corrosion.
After disassembling and thorougly cleaning the scope and battery box,
using the bright light of the illuminator and applying the new standard
at 120X magnification I was able to confirm the 2.2 (90 lppm) resolution
of the test slide.
Next, on to the Acer Scanwit. The Scanwit was tested using both
MiraFoto and Hamrick's VueScan. I had never scanned any slides with
MiraFoto and the first surprise was the discovery that MiraFoto2 does
not contain any film profile information for color slide films. I never
did ask Gary what film he used but VueScan produced a pleasing result
using one of the Ektachrome profiles. MiraFoto2 did a lousy job on the
color rendition using its generic color slide profile. The predominant
light green of the survey map was rendered as a greenish straw color.
Perhaps MiraFoto's web site has more film profile information available
but I haven't checked that yet. Maybe one of you Scanwit using list
members knows the answer. But this is a resolution discussion and I
couldn't see any resolution differences between MiraFoto and VueScan.
Using VueScan's ability to do multiple passes actually muddied the
resolution slightly at 5 passes.
The next surprise was that the Scanwit does not exhibit the same
resolution both horizontally and vertically. The horizontal resolution
is 1.6 (66 lppm) but the vertical resolution is fully two steps down at
1.25 (51 lppm). At this level one can count the lines but getting the
line fully separated along 80-900f its length requires backing down
one grade and 100 lean separation requires backing down two grades.
Post-processing the image via sharpening does improve it (by making the
lines easier to see) but the improvement is not sufficient to claim a
change in resolution to the next highest grade chart (eg: from 1.6 to
1.8). The same is true of using higher color resolution. I did the
first scans at 24 bit and another test scan at 36 bit. I didn't expect
any difference at all since I didn't think the monitor (set at 32 bit
color) was capable of showing a difference. In fact, there is a slight
improvement in visibility of the lines independent of sharpening.
However, combining both the higher color resolution and sharpening makes
a better image but still doesn't get one over the hump to the next
grade. The net result is that I'm going to stay at 24 bit color. There
isn't enough improvement in the image to justify doubling the file size
(a 36 bit scan has to be saved as a 48 bit image).
So, what's the actual resolution of the Scanwit on real world
photographs instead of horizontal and vertical test charts? I assume
that its performance is somewhere in between the horizontal and vertical
results... say, 55-60 lppm.
The next surprise was when I hauled out the old Kodak Carousel with its
102mm f/2.8 Ektanar C. This test needs to be redone under dark
conditions and at greater magnification. However, done in subdued
daylight and projecting only about 10 feet the best I could see was
about 1.1 (45 lppm). This was in the center only but was cleanly
resolved lines. Using the same standard as with the Scanwit I guess
might take the results to the next grade of 1.25 (51 lppm) and, of
course, the corner results were worse. So, the biggest surprise of all
was finding that the Scanwit appears to outdo my slide projector.
However, Gary tells me that his 102mm f/2.8 Ektanar C does 1.8 (74 lppm)
in the center and 1.4 (57 lppm) at the corner. I'll have to come back
to this one later. Just haven't had a chance to do any more yet.
So, there you have it. Some very interesting results from a $2
investment.
Your turn, Tom. What does the 4000 dpi Xerox do?
Chuck Norcutt
Woburn, Massachusetts, USA
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|