[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Rollei] Re: Rollei panorama head, and "nodal points"

> > Carlos Manuel Freaza:
> > > About the frames a bit closer, I think it depends
> > > subject and distance to subject.

> > Richard Knoppow : 
> >...Note, that since the lens moves with regard to the swivel
> point of the adaptor when the distance of focus is changed
> that the match with the nodal point will not be as good for
> close objects as for distant ones. 

The question of identifying the best rotation point for panoramic
stitching is a FAQ but the answer is not so obvious. Usually "nodal
points" are invoked, but this, pardon me Richard, is not always true.
And since we are in the Temple of Optics here, I dare to add the
following comments.

In panoramic cameras with a rotating lens drum and fixed film
w/respect to the landscape, like a Noblex, a Horizon or an Alpa
rotocamera (360degs), yes, you should rotate the lens around its rear
nodal (or principal) point. This for keeping the image almost at rest
on film, for sharpness issues while the lens rotates.

But for conventional panoramic stitching like with the Rollei TLR or
any conventional camera, where the whole camera is rotated w/respect
to the landscape, the proper rotation point to avoid parallax effects
is the __entrance pupil__. This fact is stated by Kingslake and some
other old textbooks, but is almost never explained in detail. The
basic reason why the pupils play a role here is that parallax effects
occur between out of focus image projections. An out of focus image is
built from small circles, each one being the projection of the exit
pupil on film. May be there is one object plane exactly in focus, but
when taking pictures of a flat object, you do not have to care for
parallax effects.

Rotating lens drum panoramic cameras do not respect the 'no-parallax'
condition, since the entrance pupil hardly ever coincides with the
rear nodal point. A detailed analysis of this is fascinating but would
be far off-topic here, but it suffices to say that those panoramic
cameras usually have a wide-angle lens, and that they are equivalent
to stitching many vertical strips of very narrow width, hence parallax
effects are small. Out of focus image points are slightly stretched
under the form of an oval shape, so conventional parallax effects for
out of focus images are somewhat blurred in an unconventional manner.

So the proper rotation point for panoramic stitching is the entrance
pupil. In quasi-symmetric lenses like a R-TLR lens, the entrance pupil
is located very close to the front nodal point of the lens. Another
source of confusion with nodal points !! So in order to minimise
parallax effects with a R-TLR, simply try to place the rotation point
roughly under the lens. A Rolleifix will do that pretty well. My
understanding is that some, but not all, R-TLR panoramic attachments
move the screw under the lens near the entrance pupil.

No what happens with non-symmetric lenses, like retrofocus wide angle
lenses or telephotos. 

For such lenses the entrance pupil is _not_ located at the front nodal
point, nor at the rear nodal point. Basically this depends on the
degree of asymmetry of the design, more precisely and very simply in
terms of mathematics, it is given by the amount of pupillar
magnification of the lens design. This is tabulated in manufacturers'
data sheet (I mean : good manufacturers, obviously the only ones where
RUGgers buy their lenses ;-);-);-). In German datasheet, the acronym
EP for "Eintrittspupille" is used, AP for "Austrittsspupille". EP is a
fairly obvious acronym for English-speaking readers. EP is often
measured with respect to the front lens vertex. datasheet usually
mention the diamter of the pupils, so it is easy to compute M_P and
check what follows.

Let M_p be the pupillar magnification factor equal to the ratio of
(exit pupil diameter / entrance pupil diameter), the distance between
the front nodal point H and the entrance pupil E_p is given by : 

HE_p = f*(M_p - 1) / M_p.

where f is the focal length. When M_p = 1, the entrance pupil is equal
to the front nodal point. This applies to all true symmetrical lens
designs such as classical apo-chromatic process lenses (apo-ronar,
G-claron). Most view camera lenses are quasi-symmetric, hence the
entrance pupil will be located somewhere in the front half of the
glass, close to the front nodal point.

For a retrofocus lens like Zeiss Distagons, we have M_P > 1. For such
a retrofocus, the entrance pupil will be located ahead from the front
nodal point, probably in the air in front of the lens. For a
telephoto, we often have M_p < 1. Here the entrance pupil will be
located behind the front nodal point. Since the front nodal point is
in fact behind the rear nodal point (H and H4 are crossed in a
telephoto design); it happens that the entrance pupil for telephoto is
located in the film plane !! this is actually something very hard to
figure out ; however the consequence of this is that you do not have
to add a long rail in order to properly set the right rotation point
for panoramic stitching with a telephoto ! for a 35 mm camera, the
tripod screw is close to the film plane and a small adjustment on the
slot of your tripod head will do the job !!

However aficionados of astronomical telescopes or "Telyt"
non-telephoto design of long focal lengths (without the usual negative
rear element of a telephoto), they'll have to rotate their camera
under the front part of the lens (usually a cemented doublet or
triplet), just under the glass like if it were a single lens element.
For, say, a 300 mm non-telephoto lens or a 300mm process lens used as
a long focal lens on a view camera, no way but rotating at 300mm ahead
from the film plane. Not a problem with a monorail view camera, but
this requires to use an accessory rail on a 35mm SLR.

- -- 
Emmanuel BIGLER