----- Original Message -----
From: "Randy Little" <randyslittle@xxxxxxxxx>
To: "PhotoForum educational network" <photoforum@xxxxxxxxxxxxx>
Sent: Tuesday, March 24, 2015 12:36 AM
Subject: Re: Use of Film-camera lenses with Digital
The image quality can suffer because the angle the light strikes the
sensor. With film it didn't really matter that the light was hitting at
angles. With a sensor when the light is hitting at a strong enough angle
it occludes it from the photosites farther out on the sensor. That being
said it really only matters with wide angle lens and there are fixes that
can be done in software to account for the falloff. You can also profile
the lens via a diffuse cap and then software. I use a 28mm Zeiss G lens
on my Nex 7 and it fringes purple like crazy but I fix it in post since the
lens is insanely awesome everywhere else.
Digital lens are telecentric in design which is why they are longer at the
same focal length. They use Aspherical elements to collimate the light so
it strikes the sensor straight or nearly straight on.
THis has been discussed here before at length..
I adhere to the principles of science, basically 'take no one's word' so I
tend to test and evaluate things I hear - and telecentric lenses have some
very useful properties - so I had a good look at a few and found no such
thing at all.
However I keep hearing these claims, and while some marketing material from
the digital dawn claimed this, later material claimed semi telecentric
designs or *alluded* to collimating properties, thereby making their lenses
'special' and it made them so mystical that few bothered to even ask why
they couldn't use their old lenses on these new cameras. Nowdays you do not
see these claims anywhere nnear as much. However telecentric lenses behave
totally differently to what we know as normal lenses and they do strage
things like make subjects further away appear LARGER than closer subjects -
they also have a very narrow DOF compared to 'normal' lenses, and this along
with the following has led me to conclude the claim was marketing hype.
to test the claims, I stuck some Sony and Canon 'digital' lens on a board
and had a look at the way the light behaved behind the lens - I found as
with film lenses, it formed a projection that dropped in and out of focus
exactly the same way a Canon film camera lens did. I then did my old laser
trick and shone that through the lens from the back - I used this as a focus
aid with large format cameras (there you go, a free tip from me) - basically
you point the laser at the ground glass screen causing a very bright dot of
light to form, this is projected out through the lens and .. if the lens is
out of focus, it causes the light to diverge - however when the lens *is*
in focus, an intense bright dot appears on the subject - a TTL laser focus
aid.
'digital' lenses cause the light to diverge the same as a film camera lens
does, covering the same area.
Reversing the lens and shining the light from the front of the lens,(not as
a beam focusing the camera on a dot on a sheet of card in front of the
camera projected by the laser) the light will again behave the same way as
a film camera lens, it also diverges *unless* the lens is focused correctly
on the subject, and the dot is brought into focus *at the film plane* (you
can get it in focus at other distances, but this again is exactly how a film
lens works and is the principle behind old Mamiya and Contax focusing -
shifting the film plane and not the lens).
At no time was a digital lens able to collimate the light - at no time did
any 'digital' lens behave any differently to a film lens of the same focal
length..
I did of course see different film plane / sensor plane distances, just as I
would see them differing between a Nikon and a Pentax, I also saw differing
sized image circles just as you'd expect with differing formats (just as a
5x4 has a huge image circle compared to the lens designed for 35mm
coverage). In fact I saw what appeared to be greater divergence in some
lenses designed for digi-cams as they has much shallower sensor registration
distances... (#1 notes on this below) for example the Sony E mount has a
registration distance of a mere 18mm - the Pentax Q mount is a tiny 9.2mm -
compare this to a Pentax K mount which has a whopping 45.46mm film
registration distance.. the light from the old Pentax lens must be far more
collimated than these tiny digital registration distances.
#1 - Of course the q mount has a sensor of 7.44 x 5.58 mm (9.3mm diagonal)
while the K mount has a sensor (film size) of 36 x 24mm.. (diagonal 43mm) -
but calculate the angle of divergence and it's around 60 degrees for both..
the apparently greater divergence was a presumption of mine, the
calculations proved it incorrect.. but in doing so it also showed the lenses
were projecting light out at near exactly the same angle - this blows away
any claim of special collimating properties altogether.
As a final note near all digital sensors have micro prismic arrays atop the
actual sensors - small hemispherical lenses which acto to gather the light
and focus it on the sensor. Rip apart a flatbed scanner and you'll see the
same thing - the sensors are small compared to the size of the flatbed but
they're actually still quite large (long) - the lens which focuses the
light on these sensors still has quite a wide 'angle of coverage' though ..
it's the prisms over the linear sensor however that makes sure the obliquely
angled light acually makes it to the sensor via additional light pipes.
Anyway, that is my take on it backed by evaluating the lenses and doing some
maths, finding no special properties of digital lenses - certainly nothing
even close to the properties of telecentrics - a shame really, I'd love to
have a play with one of those things - they look like a lot of fun :)
The article http://en.wikipedia.org/wiki/Telecentric_lens has an
unreferenced statement which contradicts all my findings, stating "Many
lenses that have been specially optimized for digital SLR cameras are nearly
telecentric on the image side, to avoid the vignetting and color crosstalk
that occur in color filter array-based digital image sensors with oblique
incident rays. The Four Thirds System uses this approach"
(were this true, there would be no need for microlenses above the sensors at
all) They cite no references or examples which strikes me as odd given the
number of 'digital' lenses out there in the world - the only link that is
relevant also seems to contradict the above, along with a rather nice image
of what telecentrics can do (the link is
http://www.lhup.edu/~dsimanek/3d/telecent.htm )
the relevant bit in the article also suggests 'digital' lenses do not comply
with telecentricity "Telecentric lenses tend to be larger, heavier, and more
expensive than normal lenses of similar focal length and f-number. This is
partly due to the extra components needed to achieve telecentricity, and
partly because the object or image lens elements of an object or image-space
telecentric lens must be at least as large as the largest object to be
photographed or image to be formed."
Make of this what you will - trust the manufacturers statements, trust mine,
or go test this yourself (something I'd strongly advocate) and see if your
lenses are bigger than your subjects, whether more distant subjects are
captured larger than nearer subjects, and whether you have infinite depth of
field. I suspect you'll find none of these things
k
