« Why Not Square Sensors? | Main | Canon G1X and Fujifilm X-Pro 1 »

Monday, 09 January 2012


The recent camera I've seen that actively promoted circular images was the Hobo being offered through Bostick and Sullivan. They had a few interesting images done that way.

Between the added cost of a bigger-than-necessary sensor (and the associated gotchas of wasting LCD/EVF display space) and the possibility of reduced image contrast in a lens designed to project a larger-than-necessary image circle, it seems to me that the best sensor for any given photographer is the one that most closely matches his/her desired aspect ratio, not one that requires cropping. (Substitute "most frequent" or "average" aspect ratio if you crop to arbitrary shapes).

A rectangle is as bad for photographers who shoot square as a square is for those who shoot rectangular.

I could see a nice limited niche product like the new Fuji (S/G/X/1 whatever combination of those all-too-prevalent alphanumerics) with 3 nice, fast primes. I'd be tempted. I like shooting square, but my problem shooting square is the same as shooting b&w ... I have a hard time visualizing the shot, even if I can get an EVF representation, if I *know* that I'm recording onto a color/rectangular sensor.

A few years ago, I participated in a picture-a-day challenge held on a forum and opted to shoot square shots with a normal (28 on APS-C) and convert to b/w and that made it a little easier to shoot with those parameters in mind:

- Dennis

Actually, if I remember correctly, the production process for digital sensors is similar to that of microchips, which means that each sensor starts life with all his brothers and sisters as a single large round wafer, which is then cut into small rectangular pieces. This is also one of the reasons why smaller sensors are cheaper (more so than the mere size difference would suggest): there's less leftover material after cutting out the rectangles, thus you get a higher yield from your base materials.

So technically, it would be possible to make a round sensor, but it'd be bigger than 8x10, and you'd get a lot of bad ones for each good one, or a lot of dead pixels on each one. Of course, you could cut circular sensors out of the larger circular wafer, but then you'd waste even more material than with rectangular sensors.

Just to let you in on a few of the non-goofy, non-gimmicky sides of fisheye photography:

- Plant ecology: several peers in my university biology program (mostly grad students) used a professor's 8mm Nikkor fisheye to document and measure tree canopy coverage as a percentage of visible sky from the forest floor. The 180 degree circular coverage was perfect to give horizon to horizon views. This was pre-digital (or at least, pre-full frame Nikon digital). I remember stories of difficult shots due to having to set up the camera, putting it on timer, and then trying to run away through thick bush to avoid being in the shot (depending on how high the camera was set up, I believe most of the time they could just crouch).

- Product photography involving circular items: I've loved having my Pentax fisheye zoom to get the appropriate framing and not have circular items distort despite the wide angle of view (a typical ultra wide lens has spatial distortion that the layperson believes is lens "distortion"). Think overhead shots of round tables, circular plates/place-settings, centrepieces. Harder to actually do than to describe; again, you have to be careful where you are in relation to the wide field of view.

Just thought you'd like some perspective (pun intended) on the matter.

You forgot to mention my daughter's pinhole-camera shot that appeared as a Featured Comment here back in May 2009:


We used a round sensor.

The Estes Camroc, a camera payload for Estes model rockets, took circular negatives. In fact, that's why I never got one, despite being an avid photographer and already doing my own darkroom work at the time I was doing model rocketry. Even with my own darkroom, I didn't have the capability to print a circular negative! (Durst M35 enlarger, glassless carrier.)

No comments yet either because

1) you haven't approved them yet, or

2) Your post is perfect and complete as is.

I'm going with (2).

It is certainly possible to make circular sensors with current technology, However, there's no point to doing so. Both CCD and CMOS sensors use a row-and-column architecture. Making a circular sensor would make the "A" column have three pixels in it while the "ZZZ" column in the middle of the sensor had thousands. It would still have to be read out in the same way current sensors are, processed identically, and stored in memory.

It's a lot easier to make sensors the same shape as the final image.

I've often, suspicious soul that I am, wondered if the 'proscenium arch' shape was sometimes used to disguise vignetting and poor edge definition.

I think the shape was often used to frame paintings before photography came along and another reason to use it was found.

I was one of those who wrote about circular photos yesterday. Well, I haven't taken any intentionally circular pictures today but I have taken one that I think will work well cropped to a square. Er, now I come to think of it, it might even work cropped to a circle.

Arthur Tress (from Cambria now) once did an exhibition in my hometown (San Luis Obispo) where he reversed the lens hood on his Hassy purposefully to make round images. I've met him a few times and he's definitely a most interesting character...
A link to his book 'Planets' which features the round images:

I'm relatively new to large format, but I thought you got more coverage when stopping down rather than less. If you have corners cut out of your ground glass, you can see this. If you peek through the corners at the lens, you can see the aperture. When you have stopped down far enough to see the whole aperture, you have coverage for that corner. Yes?

If I remember correctly Emmet Gowin mounted onto his 10x8 a 90mm lens meant for 5x4, which didn't cover 10x8

Um, did you add the last photo and the bit about vignetting after I made my previous comment, or am I just losing it? : o

What I was going to say this time was that I remember seeing one of those 6mm 220 degree nikon fisheyes with a camera bolted to it at a camera show years ago. I applied my eye to the viewfinder and stuck my arms out sideways. I could see both hands and elbows.

I have visions of a photographer using this lens leaning forward so as not to get their body in the photo, held up by a rope round their waist. The other end of the rope is held by an assistant who is carefully selected to be much heavier than the photographer. : )

I do like that arched framing. It's especially suitable for portraits, where it conforms to the outline of the head. If there's no corner detail, why have upper corners? The Native American teachings come to mind, something about energy getting trapped in square-cornered rooms, but flows freely in teepees and kivas.

Fisheye photos still don't appeal, though. And Wang's shot of the rockpile seems like a rookie mistake. "How many times have I told you not to get your shadow in the picture?!?" Unless he was illustrating the folk horror legend of The Hookman, which would make this a perfect photo...

The reference to the 6 mm lens reminded me of an advertising goof that recently appeared in the weekly flyer of a local big-box store. The camera ad page featured a series of zoom pictures of a skiing scene at normal, long, longer and ultra long focal lengths, accompanied by the relevant 'X' factor - 10X, 16X, 24X. But the 'normal' scene was labeled '0X'! (umm...no magnification at all?) Beyond the gaffe, though, that set me wondering - could one meaningfully talk about a lens with a focal length of 0? Never mind the engineering or the practical physics of it. Just mathematically - could one define such a lens and what would the resulting image look like? Would its field of view completely cover a hemisphere, or would it take in the whole sphere? I presume all objects would be infinitesimally small (literally '0X').

Of course you don't need to make a circular sensor to have circular images...we have the wonders of digital cropping at hand to give us an infinite variety of ways to frame pictures of our cats and computer keyboards. Brick walls are particularly fetching in oval!

By the way, that image from the Emmet Gowin book did have me thinking..."What the heck? Did they just wash ashore?"

"Um, did you add the last photo and the bit about vignetting after I made my previous comment, or am I just losing it?"

Yes, I added to this post about five times today. I mean, no, you are not losing it.


I pulled the Princeton University Press "Lewis Carroll Photographer--The Princeton University Library Albums" volume off my shelf when I read of the proscenium arch technique above.

Carroll used arches--round and elliptical and rounded off corners, all at the top of the frame. He also used complete ellipses, horizontal and vertical. This book has as well what it designates "loose prints", most of which show vignetting top and bottom; more often at the top.

I'm sorry to say--after looking at a LF ground glass often--that I didn't perceive this as a result of compensating for vignetting.

I understand that Carroll used to trim his prints, not mask them.

However some prints appear to have been vignetted by means of dodging and have less than geometric areas printed, more like a reflection of the subject.

Thanks for the information and enlightenment, always something to learn on TOP.

Sam Wang, if I recall correctly, was a Professor of Art at Clemson University for many, many years. He got his MFA at the University of Iowa under John Schulze. Hope I'm not incorrect.

Alan and Mike, you are both correct in your recollection of Sam Wang. He is a friend and neighbor, still a nice guy and great photographer, and still teaching photography in his retirement from Clemson University. I believe part of the process of his round photos was that he made the camera that made those photos. His round prints are very popular here. I have the good fortune to be working with him on some projects at our local arts center.

The round Kodak photos came from their very first cameras, which had no viewfinder; you merely pointed the camera toward the subject. The round print gave you a level horizon, no matter what. (That idea didn't last long.)
And the George Eastman House had one of those 6mm Nikon fisheyes on display, back in the '70s and '80s. It was mounted on a camera with a big prism viewfinder and you could actually look through the lens- it was just spectacular.

The comments to this entry are closed.