| Articles |


The Strange Case of the 12mm Fisheye

It's been lurking out there, largely unseen and unmentioned, for almost 40 years. It has been known by many names: Accura, Beroflex, Berolina, Panomar, Sigma, Spiratone, Universa, Upsilon, and Vemar. It's inexpensive, fun, and unlike anything else.

It's the 12mm fisheye lens.

I had never heard of this lens until recently, and piecing together its story has been a challenge. Fortunately Google has lead me to the scattered bits of information available on the web, and another clue has turned up in my own library; so I'll try to convey as much as I know about it here.

But first, we should step back a bit to gain a wider perspective (!) on this minor mystery.

Beginnings

Fisheye lenses have been with us since the glory days of 35mm SLRs, and from the beginning their basic function has been to provide a 180 degree (more or less) view of their surroundings. Though they were originally developed decades earlier for use in scientific applications, it was in the mid-1960s that they found their way into the hands of mainstream photographers who used their unique perspective to create a cornucopia of wild images. Nikon in particular led the way with their ground-breaking fisheye lenses. Other major camera makers followed, offering high quality fisheyes of two types: those that form a small, circular image on the film frame, and those that form a much larger image that occupies the full rectangular area. (I myself owned two different Minolta 16mm fisheyes back in my film days, and they were fine lenses.)

Third-party lensmakers soon joined in with their own offerings as well. Many of these past and present fisheye examples are shown here. There have also been a number of auxiliary fisheye converters introduced by third parties. These attach to the front of a camera lens and offer different types of coverage depending on the focal length of that lens (more about this subject later).

An Oddball Among Oddballs

It's generally agreed that the 12mm fisheye first appeared on the scene in 1967-68. I have the original edition of the Time/Life Library of Photography book series, and in the Special Problems volume I found a 35mm photo made by Harvey Duze in 1970 with a 12mm fisheye. I contacted Harvey, who confirmed that he used his Spiratone 12mm to take the shot. By the way, he still owns and uses the same lens with his current cameras. (If you locate the book you'll see that the photo caption gives the field of view as 150 degrees rather than 180, and that was Harvey's estimate of the true angle of coverage. My own tests suggest a slightly different figure, as we'll see later.)

More recently, Amanda Culley, a reader of my article, contacted me saying that she had purchased a boxed Sigma branded 12mm Fisheye, and thoughtfully sent me scanned images of the lens box label and user manual, both of which are very informative. (Click on those links to download PDF versions of the scans.)

It is assumed that these lenses were all designed and built by Sigma but marketed under the various other brand names mentioned above, which was a common practice. I'm not certain this is the case, but the photos I've seen of these lenses confirm that they were all visually and structurally identical, with only the markings on the front ring to distinguish them. They all looked like this:

Shown here is the large version of the lens with a T2 adapter for the Minolta/Sony AF mount, and with the ring section of the two-piece lens cap attached.

This family of fisheyes had some very distinctive traits: the odd focal length and the resulting size of the image circle; the method of aperture control; the method of mounting; and a very strange lens cap. We'll look at each of these characteristics in more detail.

Focal Lengths and Image Circles on 35mm Format

For use with 35mm film, a circular fisheye would typically have a focal length of 6mm to 8mm; and a full-frame variety would typically be 15mm to 18mm. The 12mm fisheye, however, falls in between. The image it forms on a 35mm frame is neither a complete circle nor a full rectangle. These simulations show the results that each type of lens would produce when shooting the same scene from the same camera location:

Most 8mm lenses with 22mm image circle on film

12mm lens with 33mm image circle on film

16mm lens with 44mm image circle on film

These are approximate dimensions and simulated image circles. The 180 degree fisheye image used on this page is courtesy of the University of Chicago.

I suspect that the 12mm fisheye never really saw a lot of use in the heyday of film, and I'm not sure what the advantage of its image circle was thought to be. Aside from the example I mentioned above, I don't remember seeing many published photos showing a circle with its top and bottom cut off. The photographer would have had to crop the image to make a rectangular print, and thus lose a good portion of the usable film area.

Focal Lengths and Image Circles on APS Format

In the current world of small sensor digital SLRs things are a bit different. Today's most popular DSLRs (like my Konica Minolta 5D) have a sensor size of about 24mm x 16mm, commonly referred to as APS or APS-C. To produce a proper 180 degree full circular image on these sensors, we would need a lens with a focal length of about 5mm; and to produce a full-frame image we would need a lens of about 8mm to 11mm. Only a few years ago such lenses were somewhat rare. I believe that Nikon offers a 10.5mm full-frame fisheye for their APS cameras, and Pentax has a very odd 10-17mm zoom fisheye for their own mount. Tokina also offers a similar 10-17mm zoom fisheye - but in Nikon and Canon mounts only. Recently Sigma introduced an autofocus 10mm fisheye for all major camera brands, and Samyang now markets a unique manual focus 8mm fisheye under their own brand name and several others. There are also some 8mm fisheyes from companies like Peleng that can be adapted for use on Minolta and Sony cameras. But they're manual focus, manual aperture lenses, and are not capable of covering the entire APS frame.

One other way to go is to use any of several 15-17mm fisheyes, which are easy to find. However, their image circles are far too large to retain a real fisheye effect.

So, what might we expect with all these focal lengths? Compare these APS results to those seen on a full-size 35mm frame:

Most 8mm lenses with 22mm image circle on APS

12mm lens with 33mm image circle on APS

16mm lens with 44mm image circle on APS

(Result is equal to 12mm lens on 35mm frame)

(Result is similar to 16mm lens on 35mm frame)

(Heavily cropped compared to 35mm frame)

On an APS sensor, a typical 8mm lens produces the same type of partial circle that the 12mm lens produces on a 35mm film frame. (The Samyang 8mm mentioned earlier is an exception, and it does nicely cover the APS sensor.) But notice that the 12mm lens provides a pretty close fit on an APS sensor. Its image circle is a bit too big (remember that something between 8mm and 11mm would be more correct), so some of the available image is lost. Instead of seeing 180 degrees across the corners, we can only see about 150 degrees. But hey, it could be worse! Compare this to the heavily cropped result obtained with a 16mm lens.

The oddball 12mm fisheye offers an acceptable match of image circle to APS sensor size and is also the most affordable choice out there. I've been able to snag a few of these from eBay for between $60 and $80 each, shipping included, although typical prices are in the $100 to $200 range.

By the way... if you have a DSLR with a 1.25x or 1.3x crop factor (like some Canon models) this lens should produce a nearly perfectly sized image circle for it; but I'm not able to test that capability.

Sibling Rivalry

I was very surprised to discover that there are actually two distinct variations of the 12mm fisheye! This is not readily apparent unless you happen to have both types on hand to compare, as I do. The difference lies in the assembly that holds the front elements. There is a small version and a large version. The photos below show this difference.

Both of these lenses bear the Accura brand and the small version has a lower serial number, suggesting that it was the earlier design. However, the front element of the small version appears to have more advanced coating - and that suggests the opposite. Regardless of which came first, my tests indicate that the large version is optically better. In spite of its coating, the small version provides inferior contrast and flare control. It also covers a slightly smaller field of view. Its only advantages are the smaller size and weight.

There's one more difference. Based on photos in a number of eBay auctions for the two types, it appears that the small version is always accompanied by a domed lens cap, and the large version always comes with a two-piece cylindrical cap. The rear sections of both lenses appear to be identical.

Aperture Control and Optics

There are three aperture settings: f/8, f/11, and f/16. But there is no adjustable iris. Instead the lens employs a type of 'Waterhouse stop' system - a pivoting disk with three holes of different sizes. I suppose this must have saved money on construction costs.

The maximum aperture of f/8 makes this one of the slowest fisheyes (indeed, one of the slowest 35mm lenses of any type) that I've seen. Again, that probably was a necessary decision to keep the price down. But surprisingly, this has not been much of an issue. I leave the lens at f/11 pretty much all the time, and rarely have any difficulty composing shots with it. Since there's never a need to focus, the viewfinder image doesn't have to be very bright. I just set the camera to A (or sometimes M) mode and I'm good to go.

Web sources have disagreed about how many elements and groups are present in the lens. I disassembled one of mine (the large version) and found that it has 7 elements in 5 groups. I've shown an approximation of this arrangement in the accompanying drawing. This also agrees with the specifications provided in the user manual for the Sigma lens (although, oddly, the box label itself cites only 4 groups instead of 5).

When I check for colored reflections, most surfaces appear to have simple anti-reflective coatings.

I'm sorry to say that the manufacturer has perpetrated a little white lie about the lens. Despite the markings on the front ring of some versions, it does not cover a 180 degree field of view. This is easy to confirm visually by pointing the back of the lens at a light source and looking through the front while rotating the lens until the aperture is no longer visible. With a true 180 degree lens the aperture would have to remain in view until the rim of the front element is rotated exactly 90 degrees away from your eye. Instead, the aperture becomes blocked before that happens.

 After making some careful measurements using the method above,
I estimate the actual field of view to be 165 degrees.

Coincidentally, I am aware of one other totally different (and much more expensive) 12mm fisheye lens that was made by Sea & Sea for use exclusively with the Nikonos V underwater camera. It does not function properly in air, and the company states that it covers a 167 degree field of view. Be that as it may, for simplicity I'll refer to the coverage of my lens as 180 degrees - but be aware that the figure is not strictly accurate.

Mechanics

Another unusual aspect of the lens is that it's made to use a T-mount or T2 adapter. There's really nothing wrong with this idea, since it permits the lens to be easily adapted to a wide variety of camera bodies; but I'm pretty sure there were very few if any competing fisheyes that took the same approach. Interestingly, this is one of the characteristics that makes the lens particularly attractive to today's photographers.

The small rear element protrudes a bit beyond the edge of the mount flange. For the Minolta/Sony AF mount, the distance is 2.5mm. I would think that the mirrors of most SLRs will safely clear this; but to avoid damage to the rear surface, the lens should never be rested on its mount without a rear cap attached.

The large version of the lens measures 59mm from front to rear surface; has a front element diameter of 74mm and a maximum diameter of 83mm; and weighs about 250g without a mount.

The small version of the lens measures 57mm from front to rear surface; has a front element diameter of 64mm and a maximum diameter of 73mm; and weighs about 230g without a mount.

Lens Cap and Vignetting Ring

There are screw threads surrounding the housing of the front element for attaching a lens cap. As mentioned earlier, small versions of the lens have a domed screw-on aluminum cap typical of most fisheyes. But the large versions have a two-piece aluminum ring and cap assembly. When the flat cap section is screwed onto the ring, the whole thing completely covers and protects the front element. But if you remove only the flat cap, you're left with a vignetting ring. (Is there such a term? I don't know what else to call this object.) The effect it produces on a 35mm frame is simply to mask off the outer portion of the image leaving just a 22mm 'pseudo-fisheye' circle. Compare the result to the true circular fisheye image shown earlier:

Domed cap (for small version)

Ring and flat cap assembly (for large version)

Effect of vignetting ring on a 35mm frame

The domed cap weighs about 25g and the two-piece cap assembly weighs about 75g.

Mounting and Focusing

As mentioned above, attaching the lens requires a T-mount or T2 mount for your specific camera. The difference is that a T2 mount has an inner ring with the screw threads and an outer ring with the lens bayonet, allowing rotation of the lens within the mount to achieve correct alignment on the camera. Something to watch out for when buying a lens like this with a mount already attached is that the mount or its inner ring might be frozen to the lens. It's also important that the mount is properly built and attached. An error of a fraction of a millimeter forward or back can dramatically impact the focusing ability of this lens.

The lens employs a fixed focus design, which is not unusual for a fisheye. Depth of field is tremendous, ranging from about 1 foot to infinity depending on the aperture used. Close focusing is not as good as with some other fisheyes, but there are workarounds for that. You can simply unscrew the lens from its mount slightly, which will move it forward for closer focus. Another approach requires a hair tie or a thick rubber band, although a black O-ring of the right size would probably work (and look) a little better. Slip it onto the lens as shown. Then carefully position the lens flush against your camera mount and hold it there while you shoot. This simple 'close focus ring' positions the lens forward just enough to provide perfect focus for objects that are almost touching the front element:

Hair tie, rubber band or O-ring

Wrap it around the lens mount

Lens used normally

Lens used with 'close focus ring'

What About Image Quality?

If you're looking for corner-to-corner sharpness, this is not the lens you want. But for more flexible folks, it can do quite well under the right conditions. The choice of aperture is the most important factor. Most shots at f/8 will produce an overall softness with vague halos around bright areas. This disappears at f/11 and f/16, though chromatic aberrations (CA) are always present. Surprisingly, flare seems very well controlled at the smaller apertures. I've seen sample shots posted online by others that ranged from miserable to quite good, though I don't know if those differences were due to quality control problems or user errors such as improper attachment of the mount. Could be a little of both. Below are a series of 100% center and edge crops taken from this outdoor test scene.

 

f/8 center

f/11 center

f/16 center

f/8 edge

f/11 edge

f/16 edge

 

In a different series of indoor tests below, I was surprised to see that image quality near the center can be almost as good as with my Konica Minolta 11-18mm zoom at 11mm, although the comparison breaks down as you move toward the edges. The fisheye was used at f/11, and the 11-18mm at f/8. Move your mouse over the image at the right to see the difference in perspective between the two lenses.

By the way, all of my test images were shot as Fine JPGs with sharpness and color saturation set to +1. These are my standard settings for the 5D camera.

 

12mm f/11 center

12mm f/11 edge

12mm f/11 corner

11-18mm f/8 center

11-18mm f/8 edge

11-18mm f/8 corner

Improving Results with Post-Processing

If you feel that the edge and corner performance is not up to par, bear in mind that there are ways to alter and dramatically improve these images. The problems near the edges are largely due to strong chromatic aberrations, where unwanted rainbow effects occur at high contrast boundaries. But notice how much this test scene can be improved using the inexpensive PTLens program to correct CA, plus a little extra sharpening in Photoshop. This example was shot at f/11.

Notice also that with careful choice of subject and composition it's often possible to get results that don't display the obvious fisheye distortions.

 

Center, unedited

Edge, unedited

Corner, unedited

Center, corrected and sharpened (no change)

Edge, corrected and sharpened

Corner, corrected and sharpened

One More Thing to Try

Another fun thing you can do with PTLens or similar software is to 'de-fish' the fisheye image to obtain a rectilinear perspective. The results are not as good as you'll get with a real rectilinear lens, but are still okay for many purposes. Here's another look at the office scene comparing the unedited fisheye image, a de-fished fisheye image, and the version shot with the Konica Minolta 11-18mm.

12mm original

12mm de-fished

11-18mm at 11mm

A Circular Detour

I don't want to wander off topic, but I will just mention that I also have another old, inexpensive gadget that I use to obtain circular fisheye images covering about 140 degrees. It's a front-mounted converter that attaches to the filter ring of my Konica Minolta 18-70mm lens. It can actually be used to capture full-frame images as well, if the zoom is set to about 35mm; though they don't look as good as those done with the 12mm. The lens and converter combination is also more awkward to handle than the compact 12mm fisheye. With the camera lens set at 18mm and f/8, the combination produced the image shown here.

Ambico fisheye converter

Used with 18-70mm lens at 18mm

12mm Fisheye Gallery

Let's wrap up with a few photos that show the special character of this lens. Scroll and click on a thumbnail to view a larger image:

Conclusion

So there you have the mysterious 12mm fisheye. I like this lens! It's retro-cool and easy to use, and it meets my needs as a low-cost tool for fisheye fun. I also think it's a beautiful sculptural object in itself, but perhaps that's just me.

If you can provide any further information or clarification regarding the 12mm fisheye, I would be happy to hear from you.
 

Text and images 2007-2014 Ray Lemieux (except where noted)