PDA

View Full Version : Why no retrofocus wide angles?...



Bob Fowler
24-Jun-2004, 08:12
The recent discussion about short focal length l.f. lenses has got me to thinking: Wouldn't a retrofocus design reduce the need for a center filter? I would think that because the rear cell would be farther away from the film plane, the law of inverse squares would have less of an effect on the edge brightness. OK, I'm not an optics expert so I'm not sure how covering power would be affected, but I would think that for a situation where not a lot of movements are required, a retrofocus design would (could?) be a good compromise. Opinions?

Leonard Evens
24-Jun-2004, 09:00
Most wide angle lenses that I've looked at are of slight retrofocus design. For example, the Rodenstock f/4.5 75 mm Grandagon-N has a rear flange focal length of 82 mm, which is 7 mm, longer than its focal length. I imagine the reason this is not pushed too far is that it is difficult to design an assymetric lens with good defintion over a wide field.

Sam Crater
24-Jun-2004, 09:51
I think the law of inverse squares has already been invoked by the time the light reaches the aperture. That is, it's a fundamental problem with rectilinear perspective. Imagine taking wide-angle photo of a bank of round light bulbs all in a plane parallel to your film plane. The bulbs at the edges will be larger in the image (due to rectilinear projection) but they're not any brighter, so the intensity of the light in the image has to be less.

ronald moravec
24-Jun-2004, 09:52
Not an optical engineer either, but based on 35mm experience, the lenses start to get very large. The fall in light is less, but still present.

Ernest Purdum
24-Jun-2004, 11:23
The idea of a retrofocus lens getting us away from pesky recessed lensboards is very attractive but if one were available today I would expect it to be bigger, heavier, and either have lower performance or be much more expensive than a normal wide angle of the same aperture and focal length. Since SLR's have thick bodies, they need retrofocus lenses to clear their mirrors. Their users are happy with what the lens makers provide them, but picture one of their lenses scaled up to cover 4X5 or more.

Darin Cozine
24-Jun-2004, 12:29
Do the math. Use a 75mm lens. If the light hitting the center of the film has an intensity of I, then the light hitting the long edge (not the corner) would be at roughly .6 of I. Thats a little less than 1 stop, and I doubt you would notice the difference.

So why do wide angle lenses viginette so much?

The law of inverse squares is not the only factor in causing the viginetting of lenses. If you look at an aperture of a shutter, it looks round. When you look at it from an angle, the apeture becomes more oval. This reduces the amount of light travelling through it. At a really extreme angle the aperture looks more like a pancake, which really limits the light pasing through it.

One other thing to consider is that retro-focus designs require more elements and that increases the size, weight, and expense considerably.

Paul Metcalf
24-Jun-2004, 13:16
Darin- Doesn't the front element or group essentially "straighten" the light ray path through the shutter aperture (with the rear element turning the rays back toward the appropriate place on the film plane)? This is what I recall from (limited) ray tracing study, but I could be wrong. I thought vignetting was due to physical blockage from lens rings, etc.

Paul

Kerry L. Thalmann
24-Jun-2004, 14:41
Actually, there was a line of retrofocus large format wide angle lenses made by Komura back in the 1970s and early 1980s. I know they made at least two focal lengths, a 75mm f6.3 and a 90mm f6.3. In addition, these same lenses were briefly sold by Calumet under the Caltar Pro name (not to be confused with the later Caltar Pros that were made by Schneider).

As others have theorized, they were quite large and heavy. Performance was supposed to be excellent in the center of the field, but rumored to be not as good in the corners as the Super Angulons (the "gold standard" of the day).

Flange to focal distances were about 33% longer than the focal length (about 100mm for infinity focus with the 75mm lens). This reduced the need for a bag bellows and/or a recessed board. It was also supposed to reduce the need for a center filter.

All the samples I've sen were single coated. Not sure if any were made with multicoating.

I've seen several of these lenses over the years, but never actually shot with one. I'm not a big fan of overly large, heavy lenses. Still, I am curious how the illumination fall-off of the 75mm compares to other standard 75mm wide angles. I shoot 4x5 color transparencies and have yet to use a lens shorter than 90mm that did not require (for my needs) a center filter. The 75mm f6.3 Komura SW might be such a lens (but in the end, I'd no doubt still prefer the smaller size, lighter weight, greater coverage, and better performance of my 80mm Super Symmar XL - even with the center filter).

Kerry

Darin Cozine
24-Jun-2004, 18:39
Paul, lenses like the super angulon still suffer from this. In these cases i guess you could just look at the front element from an angle and see that less light can pass through it.

Dan Fromm
24-Jun-2004, 18:57
Um, er, ah, I understand how Biogons and other lenses with tilting pupils beat cos(theta)**4 down to cos(theta)**3 or so. Also how fisheye lenses get any illumination at all in the margins of the field..

Will someone please explain how retrofocus lenses beat cos(theta)**4, if indeed they do?

Cheers,

Dan

Kerry L. Thalmann
24-Jun-2004, 22:47
Dan,

They don't "beat" the cos(theta) laws - retrofocus designs still have to obey the laws of physics. However, the value of theta changes as f' moves farther from the film plane. I'm not sure if the Komura SW lenses used a tilting entrance pupil design. Assuming they did, for the 75mm:

f' = 100mm
at the corners of a 4x5 negative (assuming a 154mm diagonal):
theta = atan(77/100) = 37.6°
cos(theta)^3 = .497

So, relative illumination at the corners would be 49.7% of the illumination in the center.

Now, the same calculation for a non-retrofocus 75mm with a tilting entrance pupil:

f' = 75mm
theta = atan (77/75) = 47.8°
cos(theta)^3 = .340

Relative corner illumination would be 34.0% of the illumination at the center of the image.

Even if the retroffocus design doesn't have a tilting entrance pupil, it would still have less fall-off in the corners than a standard 75mm design with a tilting entrance pupil - although the difference isn't nearly as significant (39.4% vs. 34.0% relative illumination) .

Kerry

Michael S. Briggs
25-Jun-2004, 00:29
At the other place this question was posted, I posted a comparison of two wide-angle Zeiss lenses, a non-retrofocus Biogon and a retrofocus Distagon. The Biogon, which uses tilting entrance pupils, has better relative illumination, when stopped down, than the Distagon. The almost symmetrical Bigon has much less distortion than the Distagon. Plus, the Biogon is simpler and hence should be and is cheaper. By analogy, it is clear why no LF retrofocus lenses are currently made.
See http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=008dqf.

Emmanuel BIGLER
25-Jun-2004, 00:59
<HTML>Some additional remarks to what has been stated. I agree 100% with everybody. Michaėl is right, however, optical design and available glasses make progress every year and the 3-rd generation of 40 mm Zeiss Distagon® for 6x6 SLR cameras (floating elements) exhibits MTF curves as impressive as the (1/2 century older) Biogon design.

35, 45 and 55 mm Rodenstock apo-grandagon lenses do exhibit a non-negligible amount of retrofocus design.

[NOW 100% OFF TOPIC ;-)] the new line of SLR lenses designed by O**s for the E-1 camera (4/3 standard) probably pushes the retrofocus design a bit further w/respect to previous 35 mm lens designs. I have followed the ads for this camera in the International press and it is a funny story. Part 1: Northern American advertising campaign, illustrated with a crazy ray tracing trying to explain how a wide angle lens will send to the sensor, rays that are almost parallel to the optical axis after crossing the lens. Part 2. Similar campaign in Germany. Ahem : in the country of Carl Friedrich Gauss, Ernst Abbe and Carl Zeiss, even amateur photographers know what a proper ray tracing is, so : exit the crazy Northern-American diagram, and the magic word 'telecentric' is introduced to explain why the lens mount diameter is so big. Part 3 : France. The marketing guys have a good idea : let's use the word 'telecentric' as a magic word of technical distinction : now we cannot ignore that O**s E-1 lenses
have something in common with a mysterious optical beast named 'telecentric lens'. In the good old days, old scientists ("opticists") with long white beards had hard times to explain how fascinating a telecentric lens can be to totally uninterested, ultra-specialized students in optical engineering who only dreamed of a 35 mm film-SLR. Now, next time you put your suit-and-tie for a formal dinner with the Upper Class of your hometown, do not forget to mention, if digital photography comes in the conversation: My lenses ? Telecentric, of course; what about yours ?"
</HTML>

Ernest Purdum
25-Jun-2004, 06:52
Ross made an early fixed magnification telephoto lens named "Telecentric" but I suppose the trademark has expired. Maybe the name was insufficiently magic.

Bob Fowler
25-Jun-2004, 10:49
My thanks to all who have responded! The math lesson was very, very interesting... :-)

Emmanuel BIGLER
28-Jun-2004, 09:43
<HTML>Too bad for you, Bob F. : here is an additional load of maths ;-) but just a simple ray tracing.

Having mentioned that when you push some ideas of the retrofocus
design to the ultimate form, you reach a kind of telecentric lens design, I should have
mentioned that a very simple ray tracing will give you an idea of what
happens. Without maths formulae.

Simply start from a huge positive lens element, at least as big as the
image size that you wish to get on your film. Place the iris not just
on the lens vertex itself but in front, close to the object focal
point, slightly less than one focal lens ahead, just behind this
focal point so that the lens is not quite telecentric but close to. Well this lens is not a t all a retrofocus, since the distance between the lens and film is equal to one focal length, but this lens+diaphragm combination shares with retrofocus design the same effect as far as light fall-off are concerned.
Let the large diameter lens itself be free of any other diaphragm.

Close the diaphragm down to a small aperture, a simple ray tracing
shows you that all incoming rays have (of course) to cross this
aperture, but the aperture beeing placed close to the front focal
point, all those rays will exit and hit the film as quasi-parallel
rays, parallel to the optical axis. Whatever their incidence angle
might be on input, they will exit almost perpendicular to film plane
!! (OK with no rear tilts ;-)

This kind of design can be considered for light fall-off issues as the ultimate form of a
retrofocus lens when the pupillar magnification becomes huge. The
entrance pupil in our example is easy to identify without running a lens
design software, it is simply the iris itself, but its image, the exit
pupil, is sent very far in front of the lens. Since the iris has a
certain diameter (otherwise no light would ever enter the lens) the
rays on exit always exhibit a residual angle, equal to the angle under
which the huge and very far-located exit pupil is seen from the film.

In this case the cos^4 theta law applies but the relevant angle is
virtually zero across the film format : this is magic ! no light loss
in the edges !! OK at least as long as rays that enter the entrance
pupil can find their way to the film, in a thick coupound lens this is
more complicated, but in principle the illuminated field of a
quasi-telecentric lens would by uniform, except of course for the
final cut-off due to the limited lens diameter.

If you want to cover a centimetric-sized silicon sensor, no problem to
get a huge lens that will cover in diameter a size as big as the image
itself. In 4"x5", at least the last lens element of an hypothetic
telecentric lens should be of 150mm in diameter.

But there would be no longer any use for a centre filter in a
telecentric wide-angle lens...



</HTML>