On Sat, Aug 29, 2009 at 1:13 PM, J.C. O'Connell<hifis...@gate.net> wrote:
>  I don't follow your logic, the higher the overall sensor resolution
> and sensor pixel densities get with time, the more the overall system
> resolution depends
> on the lens' absolute resolution AND FORMAT SIZE. If you use two lenses
> with
> same across the board resolution ( say 60 lp/mm), the larger FF sensor
> will approach
> 50% higher linear resolution (2.25X total resolution) with infinitely
> good sensors. Conversely, with infinitely good sensors, a FF lens with
> only 66% percent of the APS lens would give same overall system
> resolution.

This is only the case if the lens is the limiting factor across the
board for resolution, the field of view is the same and the pixel
density is the same. One of those factors is generally not the case in
any given situation. The biggest problem is that lens resolution
usually isn't the limiting factor except at the edges and degradation
of edge performance is unfortunately obvious in scenes with
high-frequency detail near the edges of the frame (as is quite often
the case in Landscapes, the most common application for high-MP
digital).

Note I'm not quibbling with your math, it's correct and would apply in
an ideal system.

> Now of course the sensors are not infinite resolution, but the same
> logic
> applies, all else being equal, the larger format will give higher system
> resolution
> for a given lens resolution or for same system resolution a lower
> resolution
> lens on the larger FF format can give same system resolution. This
> sounds
> like talking in circles I guess.

All things being equal, this is certainly the case.

>
> Anyway, sure if you have FF lenss that in the corners are less than 66%
> of the APS lens minimum, then nothing is to be gained. And sure, I guess
> on some mediorce or really wide angle lenses not designed for digital
> that may be occuring, but there are also many great FF lenses than have
> corner to corner performance at some apertures that exceeds 66% of the
> APS lens counterpart's resolution, and with those lenses at those
> apertures
> the net result is higher overall system resolution with FF vs APS.
>
> J.C. O'Connell (mailto:hifis...@gate.net)

The problem is that even most 'good' wide-angles from the film era are
being shown to be less than satisfactory on high-MP FF. What was
sufficient performance to be considered an excellent performer on film
is simply not enough for 20+MP digital. There are rather few
wide-angle designs, even in the 28mm range which have edge performance
suitable for high-MP FF. Once you hit around 20mm, you're starting to
run into options being countable on a single hand. Really, the only
~20mm primes that are up to high-MP digital are the Contax Distagon
21mm f2.8 T* and the second version of the Leica Elmarit-R 19mm f2.8.
Even superb performers like the Olympus 21's are running out of
resolution at ~21MP. You can, in some cases, recover the performance
in post (Canon's DPP software does an excellent job of this with
Canon's L wide zooms, as does DXOptics with any lens/camera
combination they have a profile for)

Most of those lenses which are 'good' on film do have enough centre
resolution for APS-C. But the rolloff in performance into Zone C is
the real issue. The problem is making lenses which do not show visual
degradation of edge performance vs centre performance on the two
formats because there is now enough resolving power at the edge of the
sensor for this to matter. The increased resolution across the smaller
imaging circle of APS-C is easier to achieve than maintaining
consistent resolution across the larger image circle of FF. This is
exaggerated by the fact that FF lenses must also be able to perform to
APS-C requirements due to the market bias towards APS-C (all the FF
makers also have APS-C lines which sell far more cameras and are a
major market for both APS-C and FF lenses).

A lot of this comes down to the fact that a number of the limiting
factors to resolution in 35mm film photography have been removed with
the advent of digital and that's making lenses which delivered superb
performance on film into mediocre performers on digital. This is
because film flatness and registration error for 35mm film are no
longer an issue. A sensor is nearly perfectly flat and even with
body-IS the sensor registration really cannot vary anywhere near as
much as film thickness and flatness allow.  In other words, the
rolloff on the MTF graph in zone C for 35mm lenses is becoming visible
because the factors that used to hide it are no longer an issue.

But the primary limiting factors are not the math and physics behind
resolution, but a combination of cost and market factors. You can make
lenses up to APS-C performance cheaper than you can for lenses up to
high-MP FF. And because of that, the available inexpensive FF lenses
are disappearing because they're visibly outperformed by the high-MP
FF sensors in most applications where you'd want high MP and the fact
that there really isn't a market for inexpensive FF lenses and won't
be until there's inexpensive FF bodies, which is unlikely to happen
anytime soon. So a given level of IQ costs more from FF(often
significantly so), but the absolute maximum IQ can be higher than is
achievable from APS-C.

Low-MP FF is an odd exception here because the primary applications
for it generally aren't as affected by edge performance or absolute
resolution, they tend to care more about absolute speed (AF, FPS) and
high ISO performance. But it doesn't present IQ improvements over
APS-C outside of high ISO performance.



-- 
M. Adam Maas
http://www.mawz.ca
Explorations of the City Around Us.

--
PDML Pentax-Discuss Mail List
PDML@pdml.net
http://pdml.net/mailman/listinfo/pdml_pdml.net
to UNSUBSCRIBE from the PDML, please visit the link directly above and follow 
the directions.

Reply via email to