>
> I think Chuck has a point with his comments on the diffraction-limited
> effective resolution and find his exchange with C.H. interesting.
>
I know exactly what diffraction limits are and definitely see the results in
all lens tests that I've done. However, I think what both CH and I am saying
is that we won't let math stand in the way of getting a picture not possible
by being a slave to numbers. I looked at your excellent rollover
comparison. The sharpened F22 image is awefully close to the F4 image. Had
it not been labeled, I would have been hard pressed to identify one over the
other except for the presence of the sharpened JPEG artifacts. TIFF files
would have been nearly impossible to identify.
What I find odd is the methodology for evaluating 'the best "walk
> about" wide-angle' for a full frame body.
>
No doubt. I suspected that I'd get some grief over the center-only
resolution testing. Especially since the older silver-nosed lenses
generally are better performers in the center than the later MC lenses, but
the MC lenses are more even across the entire frame. However, I have used
all but the 35-80 enough to recognize how these lenses work in the real
world for my purposes. Just by measuring the center gave me a quantifiable
reference between these lenses. If this test isn't satisfactory to all
purposes, I can understand and accept that, but it is giving me what I
wanted--a common point of reference to my specific need and that was to
profile each of the apertures for each of my lenses for center sharpness at
a common shooting distance for much of my photography.
Traditional testing checks at least center and somewhere out near the
> edge. Looking at Mike Hatem's tests of WA lenses, including several MF
> Zuikos, it's clear that the middle zone between these two may also vary
> significantly from them. The classic example is the Zuiko 18/3.5, which
> both Gary and Mike show to resolve least in the very center, improving
> as you go out. This test would show it as a relatively poor performer,
> whereas test and practical results have show it to be one of the finer
> SWA lenses ever made.
>
This is also true with my MC 24/2.8, but the gain near the fringe is not
enough to cover up the fact that the center is soft wide-open.
If I had the results of this test, I'd still be uncertain how they
> pertained to the question of comparative individual overall lens
> resolution on FF. I can see your dilemma, given the lack of a FF DSLR
> and the considerable additional time and cost of doing a film based
> test. Still, it does look suspiciously like "wasted effort" to me.
>
Well, another factor is that this test is for my specific set of lenses. To
perform a completely valid test we should probably have a sampling of at
least five lenses of each type. This way it would take into account
manufacturing variances.
> Maybe you shoudda bribed Chuck and Pat to stay an additional 1/2 day
> with dinner and a show or something and shot the tests with his 5D and
> more targets. :-)
The E-1 has greater pixel density than the 5D. 2MP per cm2 vs 1.5MP per
cm2. Granted, the 5D has a weaker AA filter which improves things quite a
bit, but since these lenses are exceeding the ability of 2MP per cm2 to
resolve to the extinction point, the 5D won't tell me anything more about
center sharpness than what I am already getting--and I'm not getting enough.
The 5D, though, would give me resolution data at the corner of a 35mm
frame.
I have previously performed these tests about three years ago using Ilford
Delta 100 with resolution charts placed in the center and all four corners.
But my lens selection has changed a little since then and I wanted to do a
comparative using the wide-angles I have now. What surprised me back was how
uniform most of the lenses were. This is why I sold the 135/3.5 and my
original 24/2.8 as they weren't. What is bothersome for this test on film is
that I still have more detail on the negative in the test charts than is
scannable at 4000dpi. I could run the enlarger to maximum height and make a
test print of a tiny region of the negative and compare that to the scanned
image and the scanned image would still be a minimum of two or three
line-pair steps in the deficit. So the digitizing process still eliminates
film from being a viable test mechanism as the extinction point of the
lenses is still beyond what the digitizing process can reveal.
I had also done E-1 images with the same lenses of the same targets at the
same distances (swapped cameras on the one tripod). The line-pairs where
dropoff occured between the scanned image and the E-1 image was close enough
to not make me concerned one way or the other. Like I said, the AA filter in
the E-1 reduced resolution to a similar point to lateral halation in the
film's emulsion. Empirical evidence from other sources (Luminous-Landscape
being one) seem to bear this out that 2MP per cm2 seems to be pretty much
the standard limit for Provia and I don't disagree in the least.
At minimum, yesterday's test revealed one thing I needed to know for
sure--what apertures to absolutely avoid with my 24/2.8 and is it better or
worse than my 35/2.8. Short answer is that it's worse than the 35/2.8
except in the middle apertures where it matched it to the best of my testing
ability.
Maybe I'll get a brick of that Efke 25 or something for full-frame
resolution testing....
AG
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|