Olympus OM enthusiasts' digital prayers have been answered ...

On 2/7/2012 14:12, Kennedy McEwen wrote:

OM lenses work just as well on digital as they did on film, which
doesn't mean telecentric lenses can't work better, but the argument was
false to begin with and was merely an attempt by Olympus to justify
their investment in cheap chips.

I have a Panasonic M43 and use *lots* of made for film lenses on it
(M42, Zeiss, Leica both M & R, M39 & others).

Some lenses, especially fast one, often don't work well at their faster
apertures (by "not well" I mean visibly less well than on a film camera).
It's hard to predict which will do well and which won't: a 85mm 1.4 only
gets good at 2.8 whereas a 35mm 1.4 is very good wide open. Also,
rangefinder lenses don't necessarily do worse than SLR lenses, just as
telephoto lenses (like the 85mm I mentioned) don't always do well.

But my definite conclusion is that, many lenses (perhaps as much as 50%)
*don't* work as well on a M43 sensor as they do on film. So there is
something going on... ;-)

On 2012-02-08 00:57:00 +0000, Kennedy McEwen said:

In article 2012020714103581999-pete3attkins@nospamntlworldcom, Pete A
writes
On 2012-02-07 13:12:37 +0000, Kennedy McEwen said:

In article , Bruce
writes
If there wasn't a problem, as you so confidently assert, why on earth
would Olympus have gone to such lengths?

To justify making small 4-turds sensors in the first place!
That's when they came out with the "telecentric is best for digital"
myth in the first place and it was shortly proven to be wrong by
measurement. The light fall-off due to non-telecentricity is *LESS* on
Olympus (and Canon FF sensors for that matter) than it was on film!

Illumination falloff for a non-telecentric lens is approximately a
cosine to the fourth power. The only way a digital sensor can suffer
_less_ falloff than film is by altering the angle of the non-central
micro-lenses.

Wrong.

3D microlens v's 2D flat film surface. A tennis ball has the same
cross section no matter what angle you view it from, while a flat sheet
of film has a cross section that is cos^2. Any 3D view of the
microlens means less than cos^4 fall-off.

Half a tennis ball clearly reduces its cross section as the angle moves
away from perpendicular.

Also, if the "image" from the micro-lens is smaller than the sensitive
area of the pixel, the illumination fall-off on a digital sensor will
*always* be less than film - by 2 of those four cosines!

That is simply trading pixel sensitivity for pixel vignetting. There's
a nice illustration on page 44 of the Leica M-Sytem Catalog (page 23 of
the PDF):

http://us.leica-camera.com/assets/file/download.php?filename=file_5691.pdf


Try measuring it, its not that difficult but too many folk would prefer
to regurgitate the dogma than find out the truth for themselves!
Alternatively, google my measurements from 5 or 6 years ago where this
was discussed to death - the Olympus dogma death!

Most (D)SLRs work adequately with most lenses because the lens mount to
sensor/film distance is relatively large. This requires wide-angle
lenses to be retrofocus designs, which is also, in effect, performing
the function of your tennis ball analogy. Indeed, a retrofocus
wide-angle shows less light falloff with both film and digital.

As an example, cos^4 of 33 degrees is 0.49, which is 1 f-stop of light
falloff. Many lenses suffer more than 1-stop of corner vignetting (when
wide-open) and this is generally considered to be acceptable. Reducing
the angle to, say, 15 degrees reduces the falloff to 13%, which is a
negligible 0.2 f-stop.

The Olympus "telecentric" argument was just an excuse for selling tiny
cheap chips. Nothing wrong with that - there is a huge market for
cheaper cameras - but they lost all credibility by trying to justify an
economic argument with fake science.

What they forgot to mention was that telecentric lenses are much more
affected by dust on the rear element :-)

In article , Bruce
wrote:

Many, if not most OM Zuiko lenses present significant problems when
used on a (Micro) Four Thirds digital sensor. The sensor design
strongly prefers telecentric lenses, where most of the light rays are
approximately perpendicular to the sensor when they hit.

the issue is nowhere near as bad as olympus would like you to believe,
and many lenses already were telecentric anyway.

But the OM Zuiko lenses were designed primarily for compactness and
light weight - small and light was the USP of the OM System - and this
coincidentally tended to produce lenses that were far from
telecentric. That didn't matter with film, which can record light
rays from all angles; all that mattered was where they hit the film,
not at what angle. But with a digital sensor, light rays striking at
oblique angles generate a much lower response from the receptors.

maybe if there aren't microlenses, but just about every sensor in the
last decade has microlenses so it's a non-issue.

The result is that many OM lenses are poor performers on digital
sensors, particularly on the small (Micro) Four Thirds sensor. They
suffer particularly from vignetting and their overall performance is
degraded compared to their performance on film.

old lenses are actually quite good, and the ones that aren't good are
because they were not very good to begin with, not because of digital.
some of them were telecentric, which blows the entire theory to bits.

isn't it strange how old nikon and pentax lenses work just fine and
produce terrific results, so much so that some of those old lenses are
actively sought after and prices are now much higher than they used to
be. even canon users like old nikon lenses (they can't use old canon
lenses due to the change in the mount).

that means that either the old olympus lenses were junk or that this
'theory' is a total crock. anyone who used old olympus lenses knows
they weren't junk, which leaves only one other possibility.

Olympus helpfully released a list of OM lenses with indications as to
which would performed well, or less well, with suggested limitations
on some in terms of lens apertures. I was so put off by the
complexity of this list, and the dire warnings it contained, that I
never even tried an OM Zuiko lens on my E-1 bodies. There was no need
because I no longer owned any OM gear and the Zuiko Digital lenses
were in any case superb.

olympus released a lot of bull**** about digital lenses that sounds
good on the surface but is nothing but bull****.

In article , Kennedy McEwen
wrote:

The Olympus "telecentric" argument was just an excuse for selling tiny
cheap chips. Nothing wrong with that - there is a huge market for
cheaper cameras - but they lost all credibility by trying to justify an
economic argument with fake science.

it was a clever ad campaign that sounds good at first, but with a
little scrutiny, it's obvious it's bull****.

In article 2012020813350381290-pete3attkins@nospamntlworldcom, Pete A
writes
On 2012-02-08 00:57:00 +0000, Kennedy McEwen said:

In article 2012020714103581999-pete3attkins@nospamntlworldcom, Pete
A writes
On 2012-02-07 13:12:37 +0000, Kennedy McEwen said:

In article , Bruce
writes
If there wasn't a problem, as you so confidently assert, why on earth
would Olympus have gone to such lengths?

To justify making small 4-turds sensors in the first place!
That's when they came out with the "telecentric is best for
digital" myth in the first place and it was shortly proven to be
wrong by measurement. The light fall-off due to non-telecentricity
is *LESS* on Olympus (and Canon FF sensors for that matter) than

Illumination falloff for a non-telecentric lens is approximately a
cosine to the fourth power. The only way a digital sensor can suffer
_less_ falloff than film is by altering the angle of the non-central
micro-lenses.
Wrong.
3D microlens v's 2D flat film surface. A tennis ball has the same
cross section no matter what angle you view it from, while a flat
sheet of film has a cross section that is cos^2. Any 3D view of the
microlens means less than cos^4 fall-off.

Half a tennis ball clearly reduces its cross section as the angle moves
away from perpendicular.

By less than a flat sheet, and 3/4 of a tennis ball reduces even less.
Hence a microlensed sensor will usually have less fall-off than film.

Also, if the "image" from the micro-lens is smaller than the
sensitive area of the pixel, the illumination fall-off on a digital
sensor will *always* be less than film - by 2 of those four cosines!

That is simply trading pixel sensitivity for pixel vignetting.
No it isn't. The purpose of those microlenses is to put the light from
the total pixel area into the sensitive area - having a smaller
sensitive area doesn't change the sensitivity of the pixel, although it
can result in lower saturation levels. Similarly, vignetting.

There's a nice illustration on page 44 of the Leica M-Sytem Catalog
(page 23 of the PDF):

A nice illustration that address a different issue, not this one. That
is showing what Leica have done to reduce light fall off even further
with very fast lenses.

Try measuring it, its not that difficult but too many folk would
prefer to regurgitate the dogma than find out the truth for
themselves! Alternatively, google my measurements from 5 or 6 years
ago where this was discussed to death - the Olympus dogma death!

Most (D)SLRs work adequately with most lenses because the lens mount to
sensor/film distance is relatively large. This requires wide-angle
lenses to be retrofocus designs, which is also, in effect, performing
the function of your tennis ball analogy. Indeed, a retrofocus
wide-angle shows less light falloff with both film and digital.

As an example, cos^4 of 33 degrees is 0.49, which is 1 f-stop of light
falloff.

And the *measured* sensitivity for 40deg incidence on a Canon 5D sensor
was as much as 94.5% of the response at perpendicular, using linear
processing of RAW data. That's a little less than 0.2 stops - and a lot
less than cos^4, which film would be expected to approximate.

Many lenses suffer more than 1-stop of corner vignetting (when
wide-open) and this is generally considered to be acceptable.

Many lenses do indeed suffer corner vignetting - they do so
independently of the sensor and this is in addition to any cos^4 or
other terms due to the sensor.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)

In article , Rol_Lei Nut
writes
On 2/7/2012 14:12, Kennedy McEwen wrote:

OM lenses work just as well on digital as they did on film, which
doesn't mean telecentric lenses can't work better, but the argument was
false to begin with and was merely an attempt by Olympus to justify
their investment in cheap chips.

I have a Panasonic M43 and use *lots* of made for film lenses on it
(M42, Zeiss, Leica both M & R, M39 & others).

Some lenses, especially fast one, often don't work well at their faster
apertures (by "not well" I mean visibly less well than on a film
camera).

Most lenses will exhibit vignetting when wide open. That has nothing
whatsoever to do with the sensor, whether digital or film, but due to
the exit pupil being truncated. You can see this happening just by
looking through the lens while tilting it - that circular aperture wide
open closes down to a "cat's eye" when significantly off axis. The
"cat's eye" is the main cause of the corner vignette. Stop the lens
down a little and tilting it has no effect on the aperture.

This was just the same with film as it is on a digital sensor - whether
you noticed it or not. Furthermore, stopping the lens down doesn't
change its telecentricity - so the very fact that you get a different
level of vignette fully open demonstrates that telecentricity isn't the
issue that Olympus claim(ed) it to be.

Lots of compact, allegedly useless on digital because they aren't
telecentric, OM lenses are used on Canon and other FF digital cameras
without problem, hence their used prices are holding up quite well.

Only this morning I was shooting with the Zuiko 18mm f/3.5, one of the
least telecentric lenses on the OM lineup due to the tiny rear pupil,
and it is extremely flat to the corners when stopped down to f/5.6 -
less than quarter a stop. Even fully open, it is only a stop down in
the corners - but I can see that looking through the lens, irrespective
of the sensor, since half the exit pupil is vignetted - nothing to do
with telecentricity. What I dislike about the Zuiko 18mm was the use of
an even number (6) of aperture blades, resulting in coarse specular
reflections and poor bokeh.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)

On 2/8/2012 17:30, Kennedy McEwen wrote:
In article , Rol_Lei Nut
writes
On 2/7/2012 14:12, Kennedy McEwen wrote:

OM lenses work just as well on digital as they did on film, which
doesn't mean telecentric lenses can't work better, but the argument was
false to begin with and was merely an attempt by Olympus to justify
their investment in cheap chips.

I have a Panasonic M43 and use *lots* of made for film lenses on it
(M42, Zeiss, Leica both M & R, M39 & others).

Some lenses, especially fast one, often don't work well at their
faster apertures (by "not well" I mean visibly less well than on a
film camera).

Most lenses will exhibit vignetting when wide open. That has nothing
whatsoever to do with the sensor, whether digital or film, but due to
the exit pupil being truncated. You can see this happening just by
looking through the lens while tilting it - that circular aperture wide
open closes down to a "cat's eye" when significantly off axis. The
"cat's eye" is the main cause of the corner vignette. Stop the lens down
a little and tilting it has no effect on the aperture.

This was just the same with film as it is on a digital sensor - whether
you noticed it or not. Furthermore, stopping the lens down doesn't
change its telecentricity - so the very fact that you get a different
level of vignette fully open demonstrates that telecentricity isn't the
issue that Olympus claim(ed) it to be.

Lots of compact, allegedly useless on digital because they aren't
telecentric, OM lenses are used on Canon and other FF digital cameras
without problem, hence their used prices are holding up quite well.

Only this morning I was shooting with the Zuiko 18mm f/3.5, one of the
least telecentric lenses on the OM lineup due to the tiny rear pupil,
and it is extremely flat to the corners when stopped down to f/5.6 -
less than quarter a stop. Even fully open, it is only a stop down in the
corners - but I can see that looking through the lens, irrespective of
the sensor, since half the exit pupil is vignetted - nothing to do with
telecentricity. What I dislike about the Zuiko 18mm was the use of an
even number (6) of aperture blades, resulting in coarse specular
reflections and poor bokeh.


Wow! Speaking about ranting and a condescending attitude!

1) I never mentioned vignetting, as vignetting per se usually wasn't an
issue when I used film lenses on my M43.

2) At least on that particular M43 sensor and camera, the "bad" lenses
at the "bad" apertures gave much softer images than on film, exposure
errors (sometimes) and a general look as if they were badly flared (no,
they were not subjected to any backlighting which would cause that, at
least on film).

3) I have (quite) a bit of experience with lens testing and I had
rigorously tested all the lenses used on film.

4) You cannot compare different sensors to each other (or to film): each
will have a different behaviour with "difficult" lenses.

5) Maybe actually read people's posts more carefully....

On 2012-02-08 16:23:43 +0000, Kennedy McEwen said:

Pete A writes
Kennedy McEwen said:

Pete A writes
[...]
Illumination falloff for a non-telecentric lens is approximately a
cosine to the fourth power. The only way a digital sensor can suffer
_less_ falloff than film is by altering the angle of the non-central
micro-lenses.
Wrong.
3D microlens v's 2D flat film surface. A tennis ball has the same
cross section no matter what angle you view it from, while a flat sheet
of film has a cross section that is cos^2. Any 3D view of the
microlens means less than cos^4 fall-off.

Half a tennis ball clearly reduces its cross section as the angle moves
away from perpendicular.

By less than a flat sheet, and 3/4 of a tennis ball reduces even less.
Hence a microlensed sensor will usually have less fall-off than film.

Yeah right - adjacent lenses will not cast shadows just as eggs in an
egg tray are each visible from all angles. Last time I checked, they
weren't.

To me, this makes sense:

http://www.olympus.co.uk/consumer/21693_7045.htm

Surely, if what you are claiming is correct then film with a rough
surface will show less light falloff with angle than one with a smooth
surface.

Also, if the "image" from the micro-lens is smaller than the sensitive
area of the pixel, the illumination fall-off on a digital sensor will
*always* be less than film - by 2 of those four cosines!

That is simply trading pixel sensitivity for pixel vignetting.
No it isn't. The purpose of those microlenses is to put the light from
the total pixel area into the sensitive area - having a smaller
sensitive area doesn't change the sensitivity of the pixel, although it
can result in lower saturation levels. Similarly, vignetting.

If the micro-lenses are so small that they do not shadow each other
then they are far too small to receive the full illuminance i.e. the
available photons per unit area. Lenses do not magically amplify the
number of photons per unit area.

There's a nice illustration on page 44 of the Leica M-Sytem Catalog
(page 23 of the PDF):

A nice illustration that address a different issue, not this one. That
is showing what Leica have done to reduce light fall off even further
with very fast lenses.

Well, slow lenses and stopped-down lenses have a low angle of incidence
so there isn't a problem in the first place. So, what precisely is the
issue you are talking about?

Try measuring it, its not that difficult but too many folk would prefer
to regurgitate the dogma than find out the truth for themselves!
Alternatively, google my measurements from 5 or 6 years ago where this
was discussed to death - the Olympus dogma death!

Most (D)SLRs work adequately with most lenses because the lens mount to
sensor/film distance is relatively large. This requires wide-angle
lenses to be retrofocus designs, which is also, in effect, performing
the function of your tennis ball analogy. Indeed, a retrofocus
wide-angle shows less light falloff with both film and digital.

As an example, cos^4 of 33 degrees is 0.49, which is 1 f-stop of light falloff.

And the *measured* sensitivity for 40deg incidence on a Canon 5D sensor
was as much as 94.5% of the response at perpendicular, using linear
processing of RAW data. That's a little less than 0.2 stops - and a
lot less than cos^4, which film would be expected to approximate.

Just goes to show that my D700 is a total pile of crap compared to a 5D.

Many lenses suffer more than 1-stop of corner vignetting (when
wide-open) and this is generally considered to be acceptable.

Many lenses do indeed suffer corner vignetting - they do so
independently of the sensor and this is in addition to any cos^4 or
other terms due to the sensor.

Of course, and these other limitations will be clearly revealed in the
lens bokeh.

"Near telecentric" does not mean "telecentric". It just means that
retrofocus (more precisely, reverse telephoto) is a lens design issue
that must be considered for small area pixels/sensels. The problem is
exacerbated by the fact that the depth of a pixel cannot be scaled: the
depth determines the probability of a photon being detected in the
material used for detection. As pixel area reduces, the depth must be
maintained therefore "tunnel vision" increases - the angle of
acceptance reduces. It is right and proper that small high-resolution
sensors have lenses designed with this factor in mind. It seems to me
that Olympus took a very bold step by attempting to define a standard
for competitors to follow. Third party lenses (and editing software)
for Nikon and Canon have to be created by reverse engineering, which
isn't exactly what the consumer wants.

I have no wish to dispel your convictions or discredit your
measurements. Quite the reverse - let's all enjoy photography despite
the endless marketing bull**** that surrounds it. Great photos usually
result from being in the right place at the right time rather than
debating the laws of physics and optics. Knowledge helps - yep, I've
ruined many a shot by choosing the wrong aperture and many more shots
by failing to pack the most suitable lenses for the subject. Hindsight
can be both a blessing and a curse.

On 2012-02-08 16:30:15 +0000, Kennedy McEwen said:

In article , Rol_Lei Nut
writes
On 2/7/2012 14:12, Kennedy McEwen wrote:

OM lenses work just as well on digital as they did on film, which
doesn't mean telecentric lenses can't work better, but the argument was
false to begin with and was merely an attempt by Olympus to justify
their investment in cheap chips.

I have a Panasonic M43 and use *lots* of made for film lenses on it
(M42, Zeiss, Leica both M & R, M39 & others).

Some lenses, especially fast one, often don't work well at their faster
apertures (by "not well" I mean visibly less well than on a film
camera).

Most lenses will exhibit vignetting when wide open. That has nothing
whatsoever to do with the sensor, whether digital or film, but due to
the exit pupil being truncated. You can see this happening just by
looking through the lens while tilting it - that circular aperture wide
open closes down to a "cat's eye" when significantly off axis. The
"cat's eye" is the main cause of the corner vignette. Stop the lens
down a little and tilting it has no effect on the aperture.

This was just the same with film as it is on a digital sensor - whether
you noticed it or not. Furthermore, stopping the lens down doesn't
change its telecentricity - so the very fact that you get a different
level of vignette fully open demonstrates that telecentricity isn't the
issue that Olympus claim(ed) it to be.

If a lens is rear telecentric then its rays are parallel i.e. the angle
of incidence is zero. Stopping it down cannot reduce the angle of
incidence any further. For a non-telecentric lens, of course stopping
it down reduces the angle of incidence thereby making it _nearer to_
rear telecentric than when it is wide-open.

Lots of compact, allegedly useless on digital because they aren't
telecentric, OM lenses are used on Canon and other FF digital cameras
without problem, hence their used prices are holding up quite well.

Only this morning I was shooting with the Zuiko 18mm f/3.5, one of the
least telecentric lenses on the OM lineup due to the tiny rear pupil,

Hang on a minute - a tiny rear exit pupil ensures that rays will not
strike the sensor/film at an oblique angle, therefore how can it be one
of the least telecentric lenses in the OM lineup?

and it is extremely flat to the corners when stopped down to f/5.6 -
less than quarter a stop. Even fully open, it is only a stop down in
the corners - but I can see that looking through the lens, irrespective
of the sensor, since half the exit pupil is vignetted - nothing to do
with telecentricity. What I dislike about the Zuiko 18mm was the use
of an even number (6) of aperture blades, resulting in coarse specular
reflections and poor bokeh.

In article 2012020818474816746-pete3attkins@nospamntlworldcom, Pete A
writes
On 2012-02-08 16:23:43 +0000, Kennedy McEwen said:

Pete A writes
Kennedy McEwen said:

Pete A writes
[...]
Illumination falloff for a non-telecentric lens is approximately a
cosine to the fourth power. The only way a digital sensor can
suffer _less_ falloff than film is by altering the angle of the
non-central micro-lenses.
Wrong.
3D microlens v's 2D flat film surface. A tennis ball has the same
cross section no matter what angle you view it from, while a flat
sheet of film has a cross section that is cos^2. Any 3D view of
the microlens means less than cos^4 fall-off.
Half a tennis ball clearly reduces its cross section as the angle
moves away from perpendicular.

By less than a flat sheet, and 3/4 of a tennis ball reduces even
less. Hence a microlensed sensor will usually have less fall-off than
film.

Yeah right - adjacent lenses will not cast shadows just as eggs in an
egg tray are each visible from all angles. Last time I checked, they
weren't.

But they *are* better than a flat surface, which is what matters in this
discussion.

To me, this makes sense:

http://www.olympus.co.uk/consumer/21693_7045.htm

Top some extent it does, but several features are exaggerated to sell
the Olympus myth. However, the edge of field example shows precisely
the effect I explained above - a reduction of the cosine law due to the
spherical surface of the microlens.

Surely, if what you are claiming is correct then film with a rough
surface will show less light falloff with angle than one with a smooth
surface.

Yes. If only there were any *films* with rough surfaces.



Also, if the "image" from the micro-lens is smaller than the
sensitive area of the pixel, the illumination fall-off on a
digital sensor will *always* be less than film - by 2 of those
four cosines!
That is simply trading pixel sensitivity for pixel vignetting.
No it isn't. The purpose of those microlenses is to put the light
from the total pixel area into the sensitive area - having a smaller
sensitive area doesn't change the sensitivity of the pixel, although
it can result in lower saturation levels. Similarly, vignetting.

If the micro-lenses are so small that they do not shadow each other
then they are far too small to receive the full illuminance i.e. the
available photons per unit area.

That isn't true. It isn't the size of the microlenses that matters, it
is their aspect ratio. ie. height to pitch. (As shown on the same
Olympus page you referenced above!) Of course, the retention of angular
response due to the microlenses eventually falls off, with the response
being much worse than a flat surface at extreme angles of incidence.
However such angles would only occur with lenses of very short focal
lengths and backworking distances - the tiny 4-turds sensor with its
consequential x2 focal length factor makes the problem worse!

There's a nice illustration on page 44 of the Leica M-Sytem Catalog
(page 23 of the PDF):

A nice illustration that address a different issue, not this one.
That is showing what Leica have done to reduce light fall off even
further with very fast lenses.

Well, slow lenses and stopped-down lenses have a low angle of incidence
so there isn't a problem in the first place.

Not so. Slow and stopped down lenses have peripheral rays with a lower
spread of angles around the principle ray than fast lenses wide open,
however that has nothing to do with the angle of the principle rays
themselves. Telecentric lenses have principle rays with zero angle of
incidence (perpendicular to the focal plane), the greater the angle of
the principle ray at the corner of the image the less telecentric the
lens is. Stopping a non-telecentric lens down increases the angle of
incidence of exactly as many peripheral rays at the corner of the image
as it decreases - the effect cancel out in all but ridiculously extreme
conditions which cannot be achieved in any 35mm SLR design.

So, what precisely is the issue you are talking about?

The angle of incidence of the principle rays to each point in the image
- departure from telecentricity, the great Satan that Olympus would have
you believe corrupts all large sensors.


Try measuring it, its not that difficult but too many folk would
prefer to regurgitate the dogma than find out the truth for
themselves! Alternatively, google my measurements from 5 or 6
years ago where this was discussed to death - the Olympus dogma death!
Most (D)SLRs work adequately with most lenses because the lens
mount to sensor/film distance is relatively large. This requires
wide-angle lenses to be retrofocus designs, which is also, in
effect, performing the function of your tennis ball analogy. Indeed,
a retrofocus wide-angle shows less light falloff with both film and digital.
As an example, cos^4 of 33 degrees is 0.49, which is 1 f-stop of
light falloff.
And the *measured* sensitivity for 40deg incidence on a Canon 5D
sensor was as much as 94.5% of the response at perpendicular, using
linear processing of RAW data. That's a little less than 0.2 stops -
and a lot less than cos^4, which film would be expected to approximate.

Just goes to show that my D700 is a total pile of crap compared to a 5D.

I have no objective measurements of a D700, let alone yours, but
subjectively it doesn't look any worse. Having said that, the 5DII has
slightly more angular light fall off - but still less than flat film.

Many lenses suffer more than 1-stop of corner vignetting (when
wide-open) and this is generally considered to be acceptable.
Many lenses do indeed suffer corner vignetting - they do so
independently of the sensor and this is in addition to any cos^4 or
other terms due to the sensor.

Of course, and these other limitations will be clearly revealed in the
lens bokeh.

"Near telecentric" does not mean "telecentric".
I didn't claim it was. The fact is that many lenses in Olympus 4-turds
range are no more telecentric than their OM equivalents despite the
smaller sensor size. Whether near telecentric or less telecentric is
all relative to sensor size.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)