Ken, I have to disagree. If the extreme CA I am seeing here (and this
is extreme - it's
been a long time since I've seen such consistent CA across the image)
were to be attributed
to the ol' excuse of "the light rays are hitting the sensor at an
angle", surely the CA
would only be prominent in the outer zones of the image, and not
towards the centre of
the image where the rays are indeed quite perpendicular to the sensor
surface. Also,
the M8 has all sorts of image quality nonsense going on, but any of
the slower 35s
(Leica, Zeiss, Voigtlander) do not show chromatic of this extent. (or
nearly, at all)
No, this is an example of a lens that stinks. I'm not saying all of
them do, maybe DPReview
got a bad copy, but the image exhibits exceptionally poor quality.
And it is difficult to design a 17mm lens for 35mm, yes, but for 4/3
you could do an exact
scaled-down (half-size) version of a great 35mm design for 35mm, and
you arrive at the
"same" end result. The twice-the-resolution-needed argument is also
not very compelling,
because the good 35mm wide angles out there arguably has more than
twice the resolution needed
for modern 20MP+ sensors, so this should be ample for a tack-sharp
half-size lens, especially
one as slow as f/2.8. The cinema industry has also had great lenses of
this focal length for decades,
and small too.
Designing a good 17mm for 4/3 is no different than designing a great
65mm for 6x7cm format
(Mamiya). And the 35's (Olympus, Leica, Zeiss, Minox) and 65's
(Mamiya) are great lenses, because
they are so easy to design.
All your arguments make the lens worse in my opinion, e.g. by bringing
the lens closer
to the sensor surface, you are freed from all of the problems of doing
a retrofocus
design, you can employ a true symmetrical / wide angle design. This
frees the optical
designer somewhat.
Your mention of no longer placing the importance on lens quality
alone, but rather as a "system"
where software corrects for lens deficiencies is worrying concept, one
I hoped would be limited
to Panasonic and their play-play lenses. Whatever you do, with a bad
lens there is information
loss than cannot be reconstructed, and this also goes counter to
Olympus' stated philosphy of
valuing lens quality above all. Also, Pentax managed just fine to make
as small, nay even smaller
pancake lenses for a much larger sensor without this level of poor
performance.
As a final comparison of why I personally think this image quality of
the Oly E-P1/17mm combo is
of dismissable quality based on what I have seen (and then I will let
the topic go, I promise),
here is a 100% crop shot on a 2004-vintage DSLR in harsh daylight
using their cheapest 35mm f/2.0
(retrofocus, much harder to design, and much faster aperture, designed
for film) lens at ISO 200
at f/7.1 with the hell sharpened out of it. Where is the noise? Where
is the CA? And Olypus is
supposed to be better at designing lenses than Canon. Especially
small, slow (f-friggin-2.8) ones.
http://fc09.deviantart.com/fs47/f/2009/168/9/5/ca_discussion_test_by_philosomatographer.jpg
I just hope this camera makes more people sell their OM kit for
reasonable prices!
On 17 Jun 2009, at 5:37 PM, Ken Norton wrote:
>>
>> It exhibits pathetic chromatic abberation, and this at f/6.3! I
>> really
>> hope this is not representative of that 17mm lens, thus far I am
>> greatly
>> unimpressed. I mean, come on, they should have AT LEAST made it an f/
>> 2.0 lens
>>
>
>
> I'd say that it's a little early to say for sure why, but I suspect
> that
> there is a bug in the processing firmware that didn't correct the
> CA. With a
> lens this close to the sensor surface you're going to get distortion
> of many
> kinds--not just CA. The in-camera processing that all manufacturers
> are
> putting in the cameras now are automatically repairing CA, color-
> fringing
> and vignetting before the RAW file is written. I suspect that this
> is just a
> case where the calibration of the processing is off and will
> probably be
> fixed in firmware version 1.1.
>
> Your criticisms are sound, but at the same time, I must challenge
> this as
> being "Old Think". In "Old Think" the manufacturers compromised size
> and
> lens-film/sensor relationship for optical precision. Now, the
> lens-sensor-processing is considered to be an unified system.
>
> To make the 17/2.8 that small and that close to the sensor means
> that the
> light-rays is no longer perpindicular to the sensor surface but
> approaching
> the surface at extreme angles. Do you remember how Olympus made a
> big deal
> about how the 4/3 lenses were specifically designed for digital by
> making
> the light rays as parallel as possible when leaving the rear
> element. If
> you can make the light rays approach the sensor direct-on instead of
> at
> extreme angles, there is no need for color and vignetting correction.
>
> However, this approach was a hardware solution to the problem. All
> manufacturers determined that hardware was preferred, but would
> still put
> the cameras at a distinct disadvantage as compared to equivalent film
> cameras because of lens-size and lens-to-sensor distance. So, software
> solutions were developed and today's processsing does a remarkable
> job of
> correcting what used to be a hardware only solution. The hardware-
> solution
> approach is exactly why Olympus 4/3 lenses are so huge!
>
> The "New Think" allows lenses to be designed more like they were
> designed
> for film cameras--allowing extreme light-ray angles to hit the sensor
> surface. Unfortunately, the sensor surface isn't flat and just the
> sensor.
> There are microlenses and UV/IR filters and AA filters. As the
> lightray
> approaches the edges of the sensor, the more distorted the lightray
> becomes
> as it has to pass through these materials of increasing thickness.
>
> What Olympus/Panasonic did to help resolve this problem is to remove
> or
> degrade the AA filter and modify the microlenses. (Kodak removed the
> AA
> filter nearly entirely from most medium-format and M8 sensors).
> Unfortunately, modifying the microlenses still affects color
> artifacts as
> well as vignetting so the in-camera processing is programmed to
> automatically correct for this.
>
> "New Think" means that since "perfection" is achieved through
> software, that
> it frees the lens designers from the constraints of size and
> distance. They
> can now make the lenses tiny. In essence, it's now marketing driving
> the
> show and saying how large the lens can be and how thick it can be.
> Obviously
> there are going to be issues. No matter how you slice it, a 17mm
> lens is
> difficult to make in any format. Had they designed it for optical
> performance, the lens would be comparable in size to an old Zuiko
> 24/2.8.
>
> 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/
>
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|
