a "normal" lens

Thanks for the informative reply.

I think you are the first person on this forum to explicitly state that
fl~=image circle is a compromise between large / expensive glass versus
aberration (do you mean by this field curvature?, or aberrations in
general?). While this is a step forward in my understanding of why the
normal lens is so good, it still doesn't have have any specific
relation to fl~=image circle, apart from coincidence? The rule doesn't
hold completely true in practice - the pentax 50mm f1.4 is fast and
excellent - and would have thought that Pentax would have been able to
produce a faster 43mm which was just as good, if it really was the
'perfect' focal length.

The links were interesting, and makes me wonder if it's possible to
produce a curved CCD to take advantage of the this field curvature
(like in the eye). Only problem with this would be it would
disadvantage telephotos...

Duncan.

Duncan Murray wrote:

I think you are the first person on this forum to explicitly state that
fl~=image circle is a compromise between large / expensive glass versus
aberration (do you mean by this field curvature?, or aberrations in
general?). While this is a step forward in my understanding of why the
normal lens is so good, it still doesn't have have any specific
relation to fl~=image circle, apart from coincidence? The rule doesn't
hold completely true in practice - the pentax 50mm f1.4 is fast and
excellent - and would have thought that Pentax would have been able to
produce a faster 43mm which was just as good, if it really was the
'perfect' focal length.

I did mention the tradeoff a while ago upthread:
http://groups.google.com/group/rec.photo.equipment.large-format/msg/5a19eb93b96b97a1

Most aberrations increase with field angle, often severely.
This makes it hard to make lenses with a large circle of
coverage, or wide field of view (depending on whether
you're reading this in r.p.e.l-f or r.p.e.35mm). If you are
content with a small field of view, you can use a long
focal length lens, which reduces the field angle needed.
Lenses used by early photographers were like this,
long and slow, with a small number of elements.
Cheap Kodak box cameras often used a simple meniscus
lens at around f/11, a bit longer than the diagonal (of
course 6x9 Brownie negs were intended to be contact
printed not enlarged).

Making the lenses faster is nice, but faster lenses have
much increased spherical aberration, and other aberrations.
That means you need to introduce more elements to the
design to get more degrees of freedom - one element
helps to cancel the aberrations of another.
If you look at a book with a history of lens designs, you can
see the lenses evolving, becoming more complex to
perform better over wider fields or at faster apertures.
New optical glasses and the development of coating
(allowing more air-glass surfaces) enabled more complex
designs. A number of good designs evolved for lenses
that cover about 50-60 degrees which is about the
f.l. ~= diagonal. The Tessar pushed the speed to 2.8,
then other designs made it faster. Large format lenses
evolved in a different direction, since speed was less
critical except for focusing.

All of these tradeoffs are a matter of degrees. Most 35mm
camera makers made (or had somebody make) a 50/1.4,
and although there are differences in the designs, I think
nearly all of them are derivatives of a Planar type. Some
sold a 55/1.4 instead, which may have been a little easier
to make because the field angles are a bit lower and it
clears the SLR mount easier, but it's still the same basic
type of lens. I'm sure all of them could have made a 45/1.4
if they wanted to. But it would cost a little more and 50mm
had become a convention.

The links were interesting, and makes me wonder if it's possible to
produce a curved CCD to take advantage of the this field curvature
(like in the eye). Only problem with this would be it would
disadvantage telephotos...

Making a curved CCD would be an enormous pain.
Also for an interchangeable lens camera, you'd have
to design all the lenses to have the same field curvature.

Telescopes designed to image over large fields often
use a weak field-flattener lens near the focal plane.
Over very large fields, for certain telescopes special jigs
were made to bend the plates to a curvature that better
approximated the field curvature. (This was for plates
that were larger than most of the CCDs in use today.)

I'm not sure if you can tell from this picture, but this Kodak
Brownie Flash Six-20 has a curved film path (the shape of
the back of the camera) that presumably partly compensates
for field curvature of its simple lens:

http://www.nwmangum.com/Kodak/BF620-1.html