I think you need to reevaluate your test target to use something more
realistic like Gary Reese did using a color, USGS survey map. I think
your USAF black & white high contrast target is giving you a false
notion of what your lenses resolve. You talk of "maxing out the sensor"
but, if you look at the diffraction limits of the lenses at various wave
lengths of light you'll see that, in a color world, it's not possible to
max out the sensor in green light at any aperture much smaller than
about f/8 on a 4/3 size sensor. Remember the diffraction limit tables
from Luminous Landscape I posted not long ago? (Table #3 near the bottom
of the page)
<http://www.luminous-landscape.com/tutorials/resolution.shtml>
At f/16 the max theoretical resolution in green light is 2MP. Let hope
you're not seeing consistent performance from f/2.8 thru f/16 as you
imply here: "The 35-80 at 35mm, essentially maxed the sensor out from
wide open to F16". In the real world of color that's not possible
unless the performance at all other apertures is only 2MP. And I
seriously doubt that conclusion.
Anyhow, time for a test target and methodology change.
As to why your exposures were way off at f/22 I can't say. It seems
highly unlikely that all of your lenses should be that much off true
f/22 but I can't argue with your observation. But I don't think it
matters. With a 4/3 sensor you should be avoiding f/22 like the plague
as your real world resolution limit there is only about 1MP.
Dr. Diffraction
Ken Norton wrote:
> Yesterday, I finally got around to a little project that I've been putting
> off for a while. I wanted to compare resolutions of various lenses at all
> F-stops. The primary reason for this little project was determining which
> of my primes was the best "walk about" wide-angle to throw on an OM body.
> To test this, I used the E-1 and a trusty USAF line-pair res chart taped to
> a door. The res chart was printed on a single 8.5x11 sheet of paper
> centered on the door, and the picture was of the entire width of the door
> with trim on each side. Even with the 24mm lens, my closest distance was
> about two meters or so.
>
> I manually metered the doorway and my starting exposure for every lens was
> 1/30 at F2.8 and I offsetted the exposure all the way down to F16 or
> F22--whichever the lens had as the maximum setting. The RAW files were
> converted with Olympus Viewer with sharpness set to zero (to minimize
> sharpening artifacts and to mimick film's lateral halation tendancies).
>
> The 24/2.8 was obviously soft wide open but would max out the sensor from
> F5.6 to F8.
> The 35/2.8 was not as soft wide open but would max out the sensor from F5.6
> to F11.
> The 35/shift was essentially maxing the sensor out from wide open to F11,
> but was still usable at F16 and even F22.
> The 35-80 at 35mm, essentially maxed the sensor out from wide open to F16.
> (Simply amazing)
> The 50 macro was soft wide open, but maxed the sensor out at nearly all
> apertures except f22.
> The 35-80 at 50mm was sharper than the 50 macro wide open, but comparable at
> all other apertures.
> The 35-80 at 80mm was sharp enough wide open and held it's own throughout
> most apertures.
> The 100/2.8 was slightly soft at F2.8, but from F5.6 to F11 was stellar. No
> idea where the topout point is.
>
> I call this a wasted effort because the limiting factor in these tests was
> the sensor. We all know the E-1 is a bit soft, but I still found it
> interesting how some of these lenses just slammed the limits of the sensor
> and kept going. I will repeat this test with a borrowed E-3 to see if I can
> get a little bit more oomph out of the data.
>
> However, it was still a success because i learned once and for all what
> apertures to shoot the 24/2.8 at and which to avoid. The 35/2.8 (an ancient
> silvernosed lens) was obviously a sharper lens than the 24/2.8 at most
> apertures, but the 35/shift slam dunked them both. The 24/2.8 is the one
> lens that has flummoxed me through the years as it can either be extremely
> sharp or extremely disappointing. I'm beginning to understand why.
>
> The 35-80 is incredibly sharp wide-open, but at the mid-apertures is pretty
> much what the other lenses in the same focal length are giving, but this
> specific sample is prone to CA and some bulging distortion. Definitely not
> an issue for people/event photography (for which it is emmenently usable,
> but for static scene pictures, probably not the best lens to grab. The 35-80
> does have more contrast than any of the other lenses which is also
> contributing to a perceived increase in sharpness.
>
> But, once I retest with a sensor with greater resolution, I'll have a better
> handle of what is happening in that zone where the E-1's sensor is maxed
> out.
>
> I didn't test the 200/4 or Tokina 100-300/4, because my working distance
> would have forced me out the back door and halfway to the
> neighbors--besides, I already have a handle on what those lenses are doing
> through other controlled tests.
>
> Where I do feel the test was a success was in extrapolating the data for
> film usage. The pixel density and AA filter characteristics of the E-1 gives
> me a good indication of the sharpness acheivable on the scanned films I'm
> using. A quarter-frame crop of a 35mm negative is giving me comparable
> sharpness to the 1/4 frame 4/3 sensor.
>
> One thing that mystified me, though. The F22 image in every case was
> anywhere from 1/4 to 3/4 a stop too bright. Either the Zuiko's F22 position
> really isn't F22 (closer to F16) or the way the digital sensor responds to
> light, there is some wierd gain at F22.
>
> 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/
|