Why do we need center filters?

[Leigh]

This is not a question about optical geometry and light fall-off. I'm familiar with optical design.

My question is that rectilinear lenses for smaller formats, even those with wider Angles of View,
don't need center filters, and in fact such filters are not even available for them.

So why are they used for LF lenses?

- Leigh

[Jim Andrada]

This is just a guess but the geometry is very different - a DSLR wide angle lens is no closer to the film plane than a regular lens so there is no extreme difference in the distance of the corner of the image or the center of the image.

Might be full of S--t but it seems reasonable at this late hour. Maybe it won't seem reasonable in the morning.

[Leigh]

Hi Jim,

Not exactly true. The from the film to the rear node is equal to the focal length when focused at infinity.

This is true for all lenses, and is in fact the definition of focal length.

So for a given AoF, the LF lens will be a longer FL, and therefore farther from the film, than a 35mm lens.

- Leigh

[Bob Salomon]

The rays that illuminate the edges and corners travel a greater distance then thos that illuminate the center. So theie is greater fall off then with regular lenses, But regular lenses also have less exposure at the edges and corners as well. just not enough to warrant correction.

If you download the relevent brochures from Rodenstock on lenses and on center filters it is all explained.

[David A. Goldfarb]

Wideangle lenses for SLRs are retrofocus designs, which inherently have less falloff of illumination, but greater distortion, which can be corrected in rectilinear lenses by adding more elements to the design or using aspheric elements. Some of the most recent wideangle lenses for LF are slightly retrofocus designs, leaving a bit more room for lens movements and reducing falloff a bit.

[rdenney]

This is not a question about optical geometry and light fall-off. I'm familiar with optical design.

My question is that rectilinear lenses for smaller formats, even those with wider Angles of View,
don't need center filters, and in fact such filters are not even available for them.

So why are they used for LF lenses?

How can this not be about optical geometry? Very short lenses on small format don't have the falloff because of their strong retrofocus design. In return, they have an assortment of geometric distortions (mustache distortion being common and annoying).

Remember that a retrofocus design is a reversed telephoto. The telephoto design uses a normal lens with a rear magnifier, and so the retrofocus design uses a front de-magnifier with a more normal lens behind it. The position of the rear node is an effective rather than an actual position in a compound design. Determining focal length is not as easy as it seems. The point is that you cannot draw a straight line of the marginal rays from subject to image surface as with a simple lens. (Consider, as an extreme example, a retrofocus fisheye. The marginal rays subtend a 180-degree angle to the subject, but only produce a circular image roughly three times the effective focal length at the image surface when focused at infinity.)

I own rectilinear lenses going down to 12mm for 24x36, 45mm for 6x7, 47mm for 6x9 and 6x12, and 65mm for 4x5. The latter two are typical biogon-type large-format lenses that are like two retrofocus lenses in opposition--this was, in fact, Schneider's breakthrough with the Super Angulon. Because of that near-symmetry, they lack any type of distortion, and their optical performance is excellent and fairly uniform across the frame. The opposing retrofocus lenses undermine the advantages of the retrofocus design, however, and that means the rear node position at the focal length is actual rather than effective, and therefore quite close to the image surface. I use center filters with these.

The former two are pronounced retrofocus designs that do not suffer from severe falloff, and center filters are not needed. But they lack corner performance that is as good as center peformance, and they suffer from various geometric distortions.

There are some lenses for small format that allegedly do not suffer from distortions, despite being strong retrofocus designs. They are also priced well outside what I can afford. (The Canon 17mm TSE lens is one example, but its corner performance relative to its center performance is still not quite as good as a later f/5.6 Super Angulon, near as I can tell.)

Rick "noting that designs targeted for digital sensors are also retrofocus designs, to increase the angle of the light rays approaching the sensor" Denney

[BetterSense]

I think the retrofocus may be part of it, but not all of it. What about wide-angle lenses for rangefinders?

My Olympus OM 28mm and 50mm lens have exactly the same physical focal length. When I use them for panoramic 'stitching', I don't even adjust the camera position; the exit pupil seems to be exactly the same distance from the film plane.

[ic-racer]

In terms of retorfocus SLR lenses, the cosine of the rays hitting the film is less. Thus less need for center filter. But there usually is falloff in most all lenses in all formats, the degree depends on the aperture and focusing distance.

[E. von Hoegh]

In terms of retorfocus SLR lenses, the cosine of the rays hitting the film is less. Thus less need for center filter. But there usually is falloff in most all lenses in all formats, the degree depends on the aperture and focusing distance.

Also the film, falloff that may be acceptable with color negative material becomes more apparent and harder to cope with on something like Vevia.

[Oren Grad]

...rectilinear lenses for smaller formats, even those with wider Angles of View,
don't need center filters, and in fact such filters are not even available for them.

Even leaving aside the CF's currently offered by Schneider and Rodenstock for their short-focal-length lenses (as short as 23mm!) designed for medium format film and/or for (small!) digital sensors, there's also the Hasselblad XPan/Fuji TX-series cameras. I don't recall about the 90, but CF's were offered at least for both the 30 and 45mm lenses.

[rdenney]

I think the retrofocus may be part of it, but not all of it. What about wide-angle lenses for rangefinders?

My Olympus OM 28mm and 50mm lens have exactly the same physical focal length. When I use them for panoramic 'stitching', I don't even adjust the camera position; the exit pupil seems to be exactly the same distance from the film plane.

"Physical focal length" is a term that will attract confusion. You are saying that the rear of the physical lens is in the same position with both lenses. That's because the 28 is a stronger retrofocus design than the 50. Necessary, of course, to prevent interference with the reflex mirror.

I have a Russian rangefinder lens of 35mm focal length that gets quite close to the film. But the rear element is quite large, too. At that focal length, which is really only a little shorter than the 43mm format diameter, the falloff is not important. The rear lens cap is deeper than it is wide.

Very wide lenses for rangefinders still have somewhat of a retrofocus design. I've looked at the 12 and 15mm Voigtlaender lenses for the Bessa rangefinders, and they were both absolutely retrofocus designs. But they were not as extreme in that design as SLR lenses. You could tell by looking, though, that they had one or more negative meniscus demagnifiers on the front and a more normal gaussian lens behind it, like most retrofocus lenses. They probably did that so that the lens would not have to project too deeply into the box and fit through that smallish opening, past the rangefinder follower and the meter device. I haven't looked at, say, the 43mm lens for the Mamiya 7, but I'll bet it's similar. None of these are symmetrical in the way large-format wide-angle lenses usually are.

There is another point, too. If a wide-angle large-format lens has abundant coverage and is used without movements, it might not need the center filter. But it might well need it when used at the extremes of its coverage. I don't use a center filter for a 90, but I might need to if I really pushed its limits for critical work with narrow film. I absolutely use the center filter for the 65--I'm always close to the edge of its image circle. That's a problem for smaller formats only when using extremely short lenses and lenses with movements, neither of which dominate the topic.

Rick "who can attest to the three-stop dropoff at the edges of the Super Angulon coverage" Denney

[Dan Fromm]

I own rectilinear lenses going down to 12mm for 24x36, 45mm for 6x7, 47mm for 6x9 and 6x12, and 65mm for 4x5. The latter two are typical biogon-type large-format lenses that are like two retrofocus lenses in opposition--this was, in fact, Schneider's breakthrough with the Super Angulon. Because of that near-symmetry, they lack any type of distortion, and their optical performance is excellent and fairly uniform across the frame.

Thread drift strikes again. Sorry, Leigh.

Um, Rick, Schneider was quite late to that party. The general design type was first used by Roosinov (wide angle Russars), later by Bertele (Wild Aviogon and Super Aviogon, Zeiss f/4.5 Biogon). As far as I know f/8 and f/5.6 Super Angulons -- see, e.g., http://www.schneiderkreuznach.com/foto_e/an_su_classic/pdf/AN_SU_68_90_R44907_1AE.PDF -- have cos^4 falloff, unlike Biogons which have cos^3. Russars and Bertele's designs take advantage of the Slussarev effect, f/8 and f/5.6 Super Angulons don't seem to.

[Jim Andrada]

Thanks - I thought it was something along these lines but as a marketing guy instead of a technical guy these last 40 years or so I only deal in high level generalities. Or put another way, as a former Chem/Physics major and software developer from the late 50"s (yes we had computers!) I have a fundamental understanding of optical principles but don't know sh-t about practical real-life details. This of course may be one of the defining differences between physicists and engineers or between, say, Harvard and MIT.

Education is what remains when you've forgotten everything you learned in college. Can't remember who said it, but then again I don't remember much anymore anyhow.

Better to have forgotten what happened than to have remembered what didn't.

[Drew Wiley]

Some of what is or is not avail in center filters is really related to marketing. Serious architectural photographers of the past were most likely to use LF, so the appropriate view
lenses had center filters made for them, and appropriately marketed. Lots of MF and 35mm
lenses have serious falloff issues too, but it simply wasn't a traditional custom to use CF's
for journalist work etc. You certainly can if needed, if you find something appropriate. Now that panoramic formats are a bit popular, filters get specified.

[Bob Salomon]

"Some of what is or is not avail in center filters is really related to marketing. "

Since I am involved in the sales of Rodenstock, Schneider and Heliopan center filters and Rodenstock and Schneider (for Linhof) lenses I will add my two cents.
Hogwash!
The vast majority of wide angle lenses from 23mm up are sold without a center filter and center filter sales are far from equaling lens sales. And that includes the sales of the 58, 72 and 90 SA XL and 80mm XL on Linhof Technoramas.

When we do sell center filters it is usually after the shooter has used the lens for what they intend doing. When a shooter orders a center filter at the same time as they buy the lens it is usually an experienced shooter who knows and appreciates the difference with and without a CF.

Modern wide angle lenses have fall off. That fall off is controllable with a center filter. Fall off is also correctable with lighting as well as with dodging and burning. What you shoot, how you shoot, what you shoot on will all play a part in how, or if, you use a center filter or another correction technique. Other lenses also fall off. Just not enough to require or need a CF. And I believe even Leica also offered a center filter for some of their 35mm lenses in the past and in the further past I think they even had a lens with a propeller to even out the results.

[Leigh]

Hi Guys,

Thanks for all the replies. Some interesting information.


Other lenses also fall off. Just not enough to require or need a CF.
This is really the thrust of my question.

For a given angle of view, the ratio of the peripheral ray length to the focal length (i.e. the cosine) will be the same.

So why does the effect appear more pronounced for LF lenses vis-a-vis those for smaller formats?

- Leigh

[rdenney]

For a given angle of view, the ratio of the peripheral ray length to the focal length (i.e. the cosine) will be the same.

Not so. There is nothing that demands that a lens have the same angle of view on both sides of the glass. That's what I meant by marginal rays not being straight (or even parallel), except through a simple lens. With complex lens designs, it's completely possible for the angle subtended by the subject in front of the lens to be different from the angle subtended by the image of that subject behind the lens.

That's why I used the fisheye as an example--it's extreme enough to easily visualize. I have a 30mm fisheye that provides a 180-degree angle of view in front of the lens. That range of subject material makes an 80mm circle on the film. If it subtended the same angle as the subject (180 degrees), the angle of light approaching the film would have to be zero, which is impossible, if the marginal rays were straight (or parallel). That's why a fisheye lens is a modern invention--it's a derivative of the retrofocus design.

In fact, a fisheye uses the same general design strategy of a retrofocus lens, with de-magnifying elements in front of something like a gaussian lens. It is just isn't designed to magnifiy the edges to provide rectilinear correction as rectilinear lenses do.

The rear of that 30mm fisheye is a 30mm filter. And that rear surface is at least 60 mm in front of the film. The crossing point of the rays is in front of that rear filter, but even if it was between the rear of the lens and the film, it could not be closer than about 40mm to the film, unless light rays bend. Thus, the light is approaching the film at most roughly 45 degrees, even though the subject material forming those rays is at right angles to the lens axis.

Example image:

There is darkening in the upper corners of this image, mostly the result of the veiling flare in the middle of the image and the natural polarization of the sky at right angles to the sun. But there is no darkening in the lower corners. The field of view of this image is 180 degrees on the diagonal. There was no center filter used (obviously), but also no correction for falloff. A center filter was not needed here.

http://www.rickdenney.com/images/dirty_devil_river_lores.jpg
Dirty Devil River, Inundated, Glen Canyon, 2001

In contrast, this image shows much less field of view (although it is still extremely wide--a 47 on 6x12), but the approximately symmetrical Super Angulon did have nearly parallel marginal rays, and the light approached the film at a much shallower angle. The Super Angulon minimizes this, by the way, by presenting a very round aperture image even at the edges of coverage, assuming a small enough aperture to eliminate mechanical vignetting. This image goes right to the edge of coverage, and the falloff was about four stops. I did not use a center filter, and applied very strong post-processing after scanning to correct it as best I could, but a significant chunk of the subject brightness range fell off the bottom of the Velvia characteristic curve and the contrast just increased. I had to leave the corners dark just to keep it from getting any weirder.

http://www.rickdenney.com/images/Niagra_cannon_scan19-20_lr.jpg
Fort Niagara, 2009

Rick "who learned this the hard way in an argument several years ago over lenses designed for use with digital sensors" Denney

[Leigh]

Not so. There is nothing that demands that a lens have the same angle of view on both sides of the glass. That's what I meant by marginal rays not being straight (or even parallel), except through a simple lens. With complex lens designs, it's completely possible for the angle subtended by the subject in front of the lens to be different from the angle subtended by the image of that subject behind the lens.
Hi Rick,

Yes, of course the subject side and the image side can be different.

The parameters I gave in the post that you quoted are "peripheral ray length" and "focal length",
both of which only apply to the image side of the lens.

Perhaps my choice of "angle of view" to describe the image-side geometry is imprecise or incorrect.

- Leigh

[Bob Salomon]

Hi Guys,

Thanks for all the replies. Some interesting information.


This is really the thrust of my question.

For a given angle of view, the ratio of the peripheral ray length to the focal length (i.e. the cosine) will be the same.

So why does the effect appear more pronounced for LF lenses vis-a-vis those for smaller formats?

- Leigh

This link will take you a brochure that has the fall-off graphs for every current Rodenstock large format lens from 23mm to 360mm.

http://www.rodenstock-photo.com/mediabase/original/e_Rodenstock_Photo_Optics_1-75__8347.pdf

This link will give an explanation of center filters

http://www.rodenstock-photo.com/mediabase/original/Filter_brochure_Center_2010_CD_Engl_10184.pdf

[Oren Grad]

This link will take you a brochure that has the fall-off graphs for every current Rodenstock large format lens from 23mm to 360mm.

http://www.rodenstock-photo.com/mediabase/original/e_Rodenstock_Photo_Optics_1-75__8347.pdf

Sorry for a tangential nitpick, but this catalog, very useful though it is, doesn't have detailed data for anywhere near the entire line-up. There's full data for at least one lens for each lens series, but not for evey focal length. The digital lens series are fairly well covered, but few of the LF film-oriented focal lengths are included.

[Leigh]

Thanks, Bob. Good info.

- Leigh

[Bob Salomon]

Sorry about that Oren. There was a brochure with all the graphs but I guess it is no longer up. At least I couldn't find it.

[Drew Wiley]

Bob - you read me wrong! I'm a strong believer in center filters, but have never seen them deliberately "marketed" or "advertised" for anything but LF or panoramic camera use. Maybe
someone did at one time - but I've never personally seen it. Old "fan" lenses included the
Hypergon. And I have put my own center filters on MF lenses - but absolutely nowhere in the literature did the mfg recommend them - even though the falloff can be significant. It's basically a different culture, with LF folks in the know, others folks typically not. And it
can be much more than just an esthetic choice - with some color neg films in particular,
change in effective exp can shift part of the image down the dye curve geometry enough
to cause a visible hue shift, and not just an illumination difference.

[jcoldslabs]

I'd be curious to know if the 43mm for the Mamiya 7 is a retrofocus design or not. I do know that the rear element of the lens is only millimeters away from the film plane, so it certainly takes advantage of the lack of a mirror box. Light fall-off isn't too bad in the corners, but it's there.

Jonathan

[Bob Salomon]

Heliopan's full line brochure details the advatages of a center filter without being format specific. But the sizes that they offer them in are common view camera sizes.

[Leigh]

Heliopan's full line brochure details the advatages of a center filter without being format specific.
Thanks, Bob. Is that available online?

- Leigh

[Bob Salomon]

Unfortunately, no. On line it is only available in german. If you need an English one just send me an email with your address.

bob@hpmarketingcorp.com

[Leigh]

Hi Bob,

The German version would work. We're only interested in graphs and tables of values, which are pretty self-explanatory.

Thanks.

- Leigh

[Bob Salomon]

www.heliopan.de

I'm not too sure that there is anything on the center filters on their web site. It may only be in the full line brochure.

[Leigh]

Thanks, Bob. I was not able to find any center filters on their product page.

- Leigh

[Bob Salomon]

We mention that they are avaiable on our web page but that is all we say.

[Leigh]

Since they're pretty lens-specific, I wonder if they're actually made-to-order, rather than stocked items.

I'm sure demand is quite low, which would result in significant inventory carrying costs.

- Leigh

[Dan Fromm]

Batch production is more likely. Setup costs have to be substantial. Carrying costs are included in price, which isn't low.

Center filters are cemented doublets. FWIW, Edmund Optics wants $395 for a "high precision" coated achromatic doublet 76.6 mm in diameter. This isn't far below the retail price of a center filter.

I just looked at B&H's site. They list five Heliopan center filters. All are 3x ND at the center, weakening to 0 at the edges. All have larger front diameters than rear, presumably to reduce mechanical vignetting. These are clearly generic, not lens-specific. "Usually ships in 7 - 14 days" There must be some inventory somewhere in the system, even if not at B&H or HP.

B&H also lists Rodenstock center filters. These seem to be somewhat lens-specific. I say somewhat because, for example, the same filter is recommended for all three Apo-Grandagons. A different filter with the same threads is recommended for 75/4.5 and 90/6.8 Grandagons.

[Oren Grad]

Center filters are cemented doublets.

There are at least two types on the market, either a single element with a vapor-deposited coating, or a cemented doublet as you say. I've never had the two types on hand to compare, but my Schneider IVA looks thin enough that it's hard to imagine it's a doublet. Rodenstock has made a fuss about the fact that the new CF's for its 23mm HR Digaron-S and 32mm HR Digaron-W are cemented doublets of gray glass, and that this is supposed to provide functional benefits:

http://www.rodenstock-photo.com/mediabase/original/Filter_brochure_Center_2010_CD_Engl_10184.pdf

[Oren Grad]

B&H also lists Rodenstock center filters. These seem to be somewhat lens-specific. I say somewhat because, for example, the same filter is recommended for all three Apo-Grandagons. A different filter with the same threads is recommended for 75/4.5 and 90/6.8 Grandagons.

The lens specificity comes not just from the size, but from the density gradient. Compare, for example, the IIIb and IIIc or the IVa and IVb in Schneider's current chart:

http://www.schneiderkreuznach.com/pdf/foto/centerfilter.pdf

[Bob Salomon]

Since they're pretty lens-specific, I wonder if they're actually made-to-order, rather than stocked items.

I'm sure demand is quite low, which would result in significant inventory carrying costs.

- Leigh

We stock them so most are available from a dealer almost immediately. In your case, in MD, usually next day since we are so close to you.

[Dan Fromm]

The lens specificity comes not just from the size, but from the density gradient.

Oren, if Rodenstock offers just one filter for all the Apo Grandagons, there's just one density gradient for the three lenses.

[Leigh]

Hi Bob,

I have no immediate need for one. I was just speculating about availability. Thanks.

- Leigh

[Bob Salomon]

There are at least two types on the market, either a single element with a vapor-deposited coating, or a cemented doublet as you say. I've never had the two types on hand to compare, but my Schneider IVA looks thin enough that it's hard to imagine it's a doublet. Rodenstock has made a fuss about the fact that the new CF's for its 23mm HR Digaron-S and 32mm HR Digaron-W are cemented doublets of gray glass, and that this is supposed to provide functional benefits:

http://www.rodenstock-photo.com/mediabase/original/Filter_brochure_Center_2010_CD_Engl_10184.pdf

Rodenstock has three types.

One for 35 to 55mm Apo Grandagon lenses
Ones for Grandagon-N lenses
And then the new version made from a gray and a clear glass combination for their digital lenses from 23 to 32mm lenses.

These center filters all have a correction of 1.5X except for the ones for the Apo Grandagon lenses. Those are 2.5X.

Current Heliopan center filters all have a factor of 1.5X. Their center density is 0.45 which is the same as the Rodenstock 1.5X CF filters.

In the past Heliopan also offered a center filter with a center density of 0.9. These are no longer in production but some dealers may still have them.

To eliminate any confusion about the numbers:
0.45 is the optical density. 0.3 = one stop. 0.6 = two stops, etc.
1.5 is the required exposure increase for a CF with 0.45 density.
3X would be the exposure factor.

Also, in order for the center filter to have any effect the lens must be stopped down at least two stops from wide open. Less then that the CF will not do anything.

[Oren Grad]

Oren, if Rodenstock offers just one filter for all the Apo Grandagons, there's just one density gradient for the three lenses.

Umm, yes, 1 = 1. Schneider is much more prolific.

I have the data sheets for the lenses. The falloff curve families for the 45 and 55 look virtually identical; the 35 curve family isn't quite the same, and is on balance a bit more severe. But the manufacturers aren't trying to achieve perfect correction anyway, just to mitigate the most severe effects, so there's wiggle room.

[Dan Fromm]

But the manufacturers aren't trying to achieve perfect correction anyway, just to mitigate the most severe effects, so there's wiggle room. Ah, yes, Heliopan.

[Oren Grad]

Ah, yes, Heliopan.

From Schneider's current CF sheet:

"To avoid too long exposure times and with consideration of the exposure latitude of the film emulsions, the vignetting of the lenses has not been fully compensated."

[David A. Goldfarb]

From Schneider's current CF sheet:

"To avoid too long exposure times and with consideration of the exposure latitude of the film emulsions, the vignetting of the lenses has not been fully compensated."

And this is why I don't really believe in the idea that center filters should really be though of as dedicated to specific lenses, apart from questions of physical size. Some of these are just 1/4 stop or 1/2 stop apart at the center. I was once shooting with a friend who had a Schneider IIIb, and I tested it against a Schneider III that I use with a few different lenses, and even comparing the transparencies side-by-side, you would be very hard pressed to tell which was the "correct" filter for which lens, because they don't correct fully. Some people seem concerned about the hypothetical possibility of exposure banding due to mismatched curves, but I have yet to see anyone produce an actual example of this. It's more of an aesthetic choice that you could only decide upon based on experience.

So personally, I have two center filters--one is dedicated to the 65/8 Super-Angulon, and it doesn't fit any other wide lenses that I have (49mm thread), and the other is a Schneider III (1.25 stops), and I use it on wide lenses that take 67mm filters (47mm SAXL, 55mm Apo-Grandagon, 75mm Apo-Grandagon, 90mm SA). If I start using the 47mm more often, maybe I'll consider the recommended 2-stop IIIc for it or a 3-stop Heliopan, more likely the latter, but meanwhile, it's not as if I can't use what I have.

[Oren Grad]

And this is why I don't really believe in the idea that center filters should really be though of as dedicated to specific lenses, apart from questions of physical size. Some of these are just 1/4 stop or 1/2 stop apart at the center. I was once shooting with a friend who had a Schneider IIIb, and I tested it against a Schneider III that I use with a few different lenses, and even comparing the transparencies side-by-side, you would be very hard pressed to tell which was the "correct" filter for which lens, because they don't correct fully. Some people seem concerned about the hypothetical possibility of exposure banding due to mismatched curves, but I have yet to see anyone produce an actual example of this. It's more of an aesthetic choice that you could only decide upon based on experience.

So personally, I have two center filters--one is dedicated to the 65/8 Super-Angulon, and it doesn't fit any other wide lenses that I have (49mm thread), and the other is a Schneider III (1.25 stops), and I use it on wide lenses that take 67mm filters (47mm SAXL, 55mm Apo-Grandagon, 75mm Apo-Grandagon, 90mm SA). If I start using the 47mm more often, maybe I'll consider the recommended 2-stop IIIc for it or a 3-stop Heliopan, more likely the latter, but meanwhile, it's not as if I can't use what I have.

In the same vein, I contacted Schneider recently to ask about the difference between the IIId and the IV, which are both 82mm, for use with a 120 SA. The representative who responded answered very sensibly - he mentioned the difference in gradient and its role in achieving a better match to the falloff characteristics of specific lenses, but added that he understood that budget can be an important consideration, and that it's perfectly reasonable to try the cheaper one (in this case, IIId) to see if the effect is sufficient for one's purposes.