On 8/2/2012 4:20 PM, Chuck Norcutt wrote:
> We agree on that but you assume it gets better than theory. I assume it
> gets worse.
>
> On 8/2/2012 2:59 PM, Ken Norton wrote:
>> The theoretical and the actual don't always match.
I think this is one of those situations where both are right and both are
wrong, at least in part because of different
understanding of what is being talked about.
DEFINITIONS
-----------------
Chuck sees the issue in rather simple, theoretical terms. increasing
diffraction will lower maximum resolution. What is
not clear in his definition is the practical extent to which a difference may
be visible. His ground of truth, so to
speak, is the calculations.
AG sees the issues more from an experiential perspective. He takes pictures at
different apertures and notices that the
images simply get sharper as he stops down from fx to fy - so diffraction must
not be a significant factor here. He also
goes on to speculate as to why that might be so when the theoretical
calculations say there should be a significant
effect. His ground of truth is in the images.
Moose is pretty much in AG's camp.* I've read Chuck's posts and calculations,
I've done test shots with my cameras and
lenses. I've found that the apertures at which my (eagle) eye can see loss of
resolution in practical, 3D subjects tends
to be one, maybe even two, stops smaller than his numbers predict. As a matter
of practice, I find Chuck's numbers
useful. I just add one stop, two in a pinch, for DOF, and am pretty sure even
one more won't be noticeable at anything
less than 100% viewing.
Remember, most of the images most of us take will be seen at sizes and/or on
media that won't resolve all that much anyway.
SYSTEM BANDWIDTH
----------------------------
A few may remember my tests years ago when I got my 5D. I took FF 5D and APS-C
300D out, set them up on a tripod on a
bright day and took images of the same subjects, with complex, 3D detail, with
the same lenses and settings.
The result was that the 5.1 MP area of the 5D sensor corresponding to APS-C
resolved more detail than the 6.3 MP 300D.
Not by much, but clearly so. Dpreview called this sort of thing this 'per pixel
resolution'. To what extent is is a
result of physical light bleeding between pixels, electronic leakage, lens
design, Bayer decoding, and/or something
else, I don't know.
System bandwidth/resolution is always less than the theoretical number given by
the number of pixels. To the extent that
is is quite a bit less in any given case, calculations showing a significant
decrease in resolution from the theoretical
number may be largely meaningless.
Suppose that the effective resolution of a 12 MP camera, including the effects
of lens aberrations, sensor design,
processing, etc. is 6 MP. (No, I'm not sure how one would meaningfully measure
this.) Calculation that diffraction
limits resolution to 11 MP at one stop, and reducing the aperture by a stop
will lower resolution further, to 7 MP, for
example, are largely meaningless. Yes, there will be an effect, but it will be
much less than the numbers stated imply,
and may simply not be discernible to the eye.
POSITIVE FACTORS
--------------------------
In the days when all these issues were first worked out, images were on film,
and viewed on paper prints. Except for
specialty areas like process camera work, using physical Unsharp Masking
techniques, there were no ways to increase edge
contrast or undo the effects of diffraction.
Today, deconvolution algorithms applied to digital images can literally undo
some the effects of various lens failings
in spreading out points and edges. Likewise, the computer descendents of
physical USM can increase contrast at
boundaries, taking detail that was inherent in the data, but not visible to the
human eye, and make it visible.
All the research and calculations for DOF and diffraction effects in
photography were done based on the ability of the
human eye to see differences in sharpness or perceived detail in prints. At
that time, every step in the process, right
through printing, lost bandwidth/sharpness.
Today, part of the whole process may, if we wish, actually increase visible
detail, reversing in post some of the losses
in capture.
OBVIOUS QUESTIONS
--------------------------
If going to f-stops lower than predicted doesn't make my images softer, my
effective resolution must be pretty low. :-(
So what's the point of going to higher MP sensors?
Simple answer : Half of 20 MPs is still greater than half of 10 MPs.
Complex answer: Refer to AG's recent posts using the analogy of audio to
explain why you need at least twice the
sampling rate needed for a specific frequency response to resolve nuances in
the overtones.
Put another way, it may be that 20 MPs are necessary in the sensor system to
get a really nice 10 MP image.
CONCLUSION
------------------
It ain't simple.
My practical experience is that post processing can, in fact, recover a
substantial part of the visible detail loss from
various limitations of the capture process.
" Sacrifice and recovery of sharpness in the pursuit of DOF is no vice."
Diffracted Moose
* Wearing his "I Agree with the other guy with a Pony Tail" T-shirt.
--
What if the Hokey Pokey *IS* what it's all about?
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|