By Eduardo Cervantes
I was interested in Mike's article the other day called "Why Not Square Sensors?" I have advocated before for square sensors. In the case of a full-frame DSLR, an ideal square sensor would be 36x36 mm in size, but a proportional conversion could implemented in any digital camera known to man.
The reason for a square sensor is to go from vertical to horizontal at almost the speed of thought. It would allow the operator to hold the camera always in the same way, preserving the controls always in the same place and always keeping the flash over the center of the lens. No fancy, tiring and distracting elbow movement either. Superior to working with a vertical grip too, I believe.
This proposed system must be mirrorless. The main advantage is that the entire existing line of lenses in any camera system could be used without any restriction or limitation. Only three things are needed: A square shutter curtain to cover the entire square sensor in cameras with shutters, a square LCD and a square EVF. Perhaps, the complete shutter could be discharged with the use of an electronic shutter. They are already here.
The photographer could go from vertical to horizontal and to square at the soft inconspicuous touch of a dedicated button to obtain the desired framing in both the LCD or EVF. He or she could go from a 1:1 to a 2:3 aspect ratio and anything in between with the same camera. Whatever suits the subject better. One other huge bonus would be that when shooting Raw, the in-camera cropping would not be final. Re-framing and re-orienting the picture could be done during Raw developing always keeping the maximum sensor area in regard to the image circle.
The purpose of this arrangement is appointed for speed and comfort when shooting horizontals and verticals as well as for the optimization of the image circle in relation to subject kind and intention. It is not meant for the entire square sensor to be covered with the lens image. Suffice to say, the efective image would be kept within the projected circle of the lens.
For an existing lens system ready for a 24x36 sensor, these are the rounded figures:
2:3 aspect ratio: 24x36 = 864 square millimeters
3:4 aspect ratio: 25.5x34 = 867 sq. mm.
4:5 aspect ratio: 26.5x33 = 874 sq. mm.
1:1 aspect ratio: 30x30 = 900 sq. mm.
One more thing: There are some camera systems with a few perpective control lenses also known as shift lenses. As we all know, these lenses have bigger circles in order to be able to shift the lens without running out of image.
Nikon and Canon have some superb T/S glass in their lens lines. In the particular case of Canon, the new T/S-E II perspective control lenses have a usable circle of 67.2 mm. The circle needed to cover the entire 36x36mm sensor is 51mm. By using this kind of lenses with a square sensor we would effectively be converting a full-framer into a digital medium format camera and still have some millimeters left for shift.
Hasselblad's V camera first digital back had a 36x36 sensor. The image area of a 36x36 sensor is 1296 square millimeters in area, just a few millimeters shy of the Leica S2 with 1350 square millimeters in its 30x45 sensor.
The extra cost of the square sensor, especially as the cost of sensors continues to fall, would be within reason when taking into consideration all these benefits.
Eduardo
Mexican photographer (and TOP reader) Eduardo Cervantes studied photography in the United States and now owns and operates an all-digital professional studio in his native Merida, Yucatan.
Send this post to a friend
Please help support TOP by patronizing our sponsors B&H Photo and Amazon
Note: Links in this post may be to our affiliates; sales through affiliate links may benefit this site. More...
Original contents copyright 2011 by Michael C. Johnston and/or the bylined author. All Rights Reserved.
Featured Comment by Francisco Cubas: "A small world isn't? My first photography teacher was Edmundo Segura, in Villahermosa, and he used to talk about Eduardo Cervantes, who lived for some years in the city. He'll be glad to read this."
Lets do some silly maths using the Blue Rectangle and Yellow Square example (same diagonal therefore same lens, assuming no internal baffles etc.).
Yellow square produces 900 sq. mm, with 10% of images uncropped. The remaining 90% get cropped into 2:3 aspect ratio. That means 90% of final images are 600 sq. mm and 10% of images are 900 sq. mm for a total average of 630 sq. mm for the Yellow Square format.
Blue Rectangle produces 864 sq. mm, with 10% of images cropped into squares. The remaining 90% go uncropped at 2:3 aspect ratio. That means 90% of final images are 864 sq. mm and 10% of images are 576 sq. mm for a total average of 835 sq. mm for the Blue Rectangle format.
So, actually, on average and with very silly assumptions, the Blue Rectangle has a much higher average surface area utilization than the Yellow Square. In other words, the Square format only makes sense for those photographers that will output an overwhelming majority of images in the Square format, which is almost none. That makes it a very small niche market and therefore probably not very profitable.
The most efficient solution (cheapest, whatever), is for those that insist on outputting Square images to adapt existing Rectangular formats to their way of working.
Having said that, it would be totally awesome to see a Square format camera being marketed by a major player. Samlympus?
Posted by: beuler | Friday, 13 January 2012 at 10:11 AM
You can also squeeze a paroramic ratio like 16:9 in there quite nicely, as my GH2 does, using the extra side pixels. I think it's a good idea, and I suspect one companies have considered. Perhaps it has more "camera geek" appeal than general marketing appeal.
Posted by: John Krumm | Friday, 13 January 2012 at 10:13 AM
All due respect, it si not so simple. I believe you grossly underestimate the full scope of the issue.
While I, in concept, would like square sensors, I believe your sizing to be arbitrary and not contemporary.
While major makers have continued to build, and in some cases, design, new full frame lenses they are becoming totally professionally oriented and "full frame 35mm" cameras are Pro oriented.
That said many pros have gone to variations of APC.
I also belive that there is an over simplification of shutters etc.
Should a maker choose to do a square variant of any existing format it would be foolish to build a system around it when the user could simply crop square from the existing formats.
Without a totally new body design there is also no compelling reason to make a square variant for the same reason.
From a makers perspective the design and fabrication cost would likely not be recovered and there are existing "medium format" digitals readily available.
You seem to have missed the requirement of new designs for electronics supporting the sensors as well.
So it seems your square is relatively simple actually requires special lenses, new designs, new fabrication; a significant investment for an extremely limited market.
So despite my personal love of square and the hasselblads sitting on my shelf unused unable to justify the cost of a medium format back I cannot support any of your logic.
My suggestion? The same as you support for any portrait or lanscape picture from a square sensor. CROP.
Kindly
Posted by: bill vann | Friday, 13 January 2012 at 10:39 AM
Bravo. And I hope someone puts that sensor into a twin-lens reflex body. I'm rarely happier than when I'm shooting with a Rollei, a Mamiya, or the Ricohflex that my dad brought home from the Korean War. They're all a delight to use, and the Rollei and Mamiya produce beautiful negatives. The Ricoh's are, umm, atmospheric.
Great cameras all, but I'll gladly give up scanning, if someone markets a well made digi-TLR.
Posted by: John | Friday, 13 January 2012 at 10:58 AM
Square format is a good idea. Since the full 35mm format is 36mm wide, most full frame lenses should work, and if corner coverage is an issue, well, using a 33mm square wouldn't lose much, and would permit coverage by more lenses. Square is a reason Hasselblad and other 6x6 has been so popular, although for many 35mm users, camera size is an issue. The few times I have been able to use a Hassy, I found that it only took a few minutes to adapt to the camera size and shape. However, I don't see any reason that a properly designed mirror shouldn't work well in a square format digital. Most digital SLRs have a 'mirror up' mode already for the few times mirror shake is an issue.
Posted by: Richard Newman | Friday, 13 January 2012 at 11:15 AM
Frankly, I think cycling through aspect ratios and orientations using the "soft inconspicuous touch of a dedicated button" would be about 10 times as distracting and time consuming as simply rotating the camera 90 degrees.
If a square sensor camera were to be made, I would skip all the aspect ratio/orientation shenanigans and just crop in post. It amounts to the same thing, and allows the photographer to focus on the subject instead of buttons and settings. (This is particularly important in a dynamic setting where the subject is constantly changing; less so in a studio situation.)
Posted by: Ed Hawco | Friday, 13 January 2012 at 11:17 AM
One more benefit to having a square or round sensor, post-shot leveling. Or even in-camera horizon correction. I know that I use the tilt correction feature in Picasa more than I care to admit.
Posted by: David Owensby | Friday, 13 January 2012 at 11:34 AM
I believe the Panasonic GH2 already does something similar with its multi-aspect ratio sensor, although only in the horizontal orientation as it is still a rectangular sensor. DPreview has a demonstration graphic here:
http://www.dpreview.com/reviews/PanasonicDMCGH2/
Posted by: Cary Seipp | Friday, 13 January 2012 at 12:05 PM
Eduardo and Mike,
I'm so glad the conversation has gone in this direction! Since Mike's article on square formats, I've been looking through some of my favorite portraits from my TLR. After some thought, I did a little bit of math to see if I could replicate the relaxed angle of view and shallow depth of field of a 80mm f/3.5 on 6x6. Here is what I learned:
An 80mm f/3.5 lens has a field of view of about 4 feet wide by high, at a subject distance of six feet (spacious for the seated portraits that I favor), and a depth of field* of about 6 and 1/2 inches.
On 4x5, an equivalent field of view, and depth of field, at the same distance, for a square print can be produced with a 135mm f/5.6 lens. Blurriness at infinity should be very similar, given that the physical aperture is 24mm wide at 5.6, within a millimeter and a half of the 80 f/3.5.
On full frame, I discovered that finding a square format equivalent was more of a problem: if you merely needed the same width, a 3:2 ratio gives you 51mm as an equivalent, but if you need it to be square, and shallow, you have to go out to a 35mm f/1.4. (This actually provided some insight as to why Mike is an aficionado of 50mm and 35mm normal lenses.) I was immediately troubled, because I know that there are few (affordable) equivalent fast lenses for APS-C, and I didn't see how to get there for 4/3ds at all, given the shallow depth of field constraint.
The math predicts that a 24mm f/0.7 lens, would have the right field of view, but would be impractically fast. Assuming a f/1.4 lens, depth of field doubles to about a foot. That's not exactly favorable, and neither are the prices for fast 24mm lenses. Samyang does make a nice manual focus 24mm f/1.4, and Sigma's f/1.8 is huge, albeit, workable at the price of a modest hit to shallow depth of field. A compromise solution that retains the "right" depth of field would be a 30mm f/1.4, which loses you six inches on either side of the subject, while gaining you an extra three on the other two sides. Not ideal, but that lost six inches around the edge were probably partially accounted for in the metal mask around the edges of the frame on my Yashica TLR. In other words, probably not far from what I would have actually seen in the viewfinder.
Oh, and Micro Four-Thirds? And everybody's favorite lens, the 20mm f/1.7? Turns out that the square-er proportions of 4/3 help quite a bit: a subject distance of 6 feet gets you almost exactly four feet on a side, and one foot 2 and 1/2 inches of depth of field. Hardly worse than the best case for APS-C, and hey, it's affordable and autofocus. Blur at infinity should be noticeably different however, given a physical aperture size of only 11mm instead of 24mm. Apparently I'll have to do more math (and testing!) to see how close I can get to what I used to see.
Mike, I appreciate your patience with my technical mumblings about depth of field. And everyone, please feel free to correct me if I've mangled the math.
Will
Posted by: Will Frostmill | Friday, 13 January 2012 at 12:44 PM
I'v often wondered this as well, but I'v assumed that yields/cost's make this feature potentially very expensive, at least if you look at the manufacturing cost difference of APSC and 35mm sensors (over 20x more expensive if memory serves).
However there have been rumours that Canon are looking to get into the MF market if they lose market share by not having a high mega-pixel studio camera, so maybe this could be a nice compromise in that area, and allow existing lenses to be compatible, allowing an easier transition.
Posted by: Rhys Wheatley | Friday, 13 January 2012 at 01:03 PM
Interesting concept. Lens hoods would need to be redone though!
Posted by: Christer | Friday, 13 January 2012 at 01:09 PM
I get the appeal but practically speaking:
1. Many existing lenses have internal masks & hoods that are rectangular or optimized for horizontal capture (like all those petal hoods)
2. Coverage is expensive in many ways, whether it's the recording medium (sensor size), the real-estate a mechanical shutter must cover, the size/shape of the LCD for both review or in a EVF. And most existing display targets are rectangular -- photo/printer paper, picture frames, computer monitors, TVs, smartphones, etc. Allowing the artist to choose one dimension or the other to maximize is a choice.
I like the compromise that Panasonic has included in some cameras -- a slightly over-sized, slightly over-square sensor that allows a little less waste and a little more optimization when choosing among different rectangular formats (including 16:9 which Eduardo didn't include in his concept) -- but without going full-monty with supersized square sensor that would probably cost considerably more and cause other problems like lower x-sync speed with focal plane shutter.
Posted by: 57thStIncident | Friday, 13 January 2012 at 01:44 PM
Cary Seipp's correct that this feature exists in the GH2, and in fact it appeared before that. The GH1 had it, and I suppose it's possible that the G1 (or one of the other early Panny Micro 4/3rds cameras) had it as well.
I do like the ability to change the aspect ratio in-camera with the GH1 to help me frame the shot the way I want. While you can of course always crop to different shapes in post, I'd much rather be able to see it as I'm shooting.
How about a square sensor camera with a rotating switch to control the aspect ratio and orientation? Switch at 0 degrees is 2:3 vertical, 22.5 degrees is 3:4 vertical, 45 degrees is square, 67.5 degrees is 3:4 horizontal, and 90 degrees is 3:4 horizontal. That would allow you to always keep the camera in one orientation but give easy tactile feedback (position of the lever on the rotating switch) as to what aspect ratio/orientation you have selected.
Other comments make valid points about some disadvantages of going to asquare sensor, but it would be nice if at least one manufacturer offered it as a feature. So many cameras are all trying to occupy the same niches that this could be one way to truly differentiate a camera line.
Posted by: Andre | Friday, 13 January 2012 at 02:18 PM
I second the comment about the TLR.
But in all seriousness, a mirrorless camera with a square sensor, whether it was multi-aspect-ratio or not, would be very useful. Even more useful would be such a camera, with about the same volume (though not necessarily the same specific dimensions) as my EP-2, but with a live-view LCD that tilted upward as much as 90 degrees.
Marketed by Panasonic/Samsung/Sony, it might look like The Future(!). Or marketed by Fuji, it might look like a little Hassy.
Either way, it would make inconspicuous shooting really convenient, and would be a boon to those of us who do a lot of cropping anyway.
Posted by: Adam Z. | Friday, 13 January 2012 at 02:26 PM
The best part of the TLR idea (awesome, John!) is that with an EVF the upper lens could be totally fake ;)
The top half of the body could be filled with battery, accurately reproducing that insanely top-heavy feel of the Mamiya C3xx series!
Posted by: Andrew Molitor | Friday, 13 January 2012 at 02:33 PM
1. Regarding the desirability of flash orientation above the lens. It is a key issue for me when using a flash. So much so, that I will shoot horizontally with the camera, even though I know I will be cropping vertically. Yes I know I can use a device that swings the flash 90 degrees, to compensate for changing camera orientation, but what a bother and distraction. With enough resolution, I don't need a square sensor, just enough vertical pixels. With my camera, I now have enough pixels.
2. Medium format initially used a square, I suspect first with the Rollei twin lens. Pretty hard to turn a twin lens sideways. When medium format SLRs started, they initially copied the square. There was enough resolution versus 35mm to allow the photographer to throw away (crop) the film (sensor) But MF SLRs then moved to 645 and rotating 6x7.
3. Large Format never used a square, as the film could be rotated.
4. I would argue that the first digital camera that had the flexibility to be any format it wanted, was the first digital camera that had the vision to realize it was no longer tied to a mass produced film size limitation.
Interestingly it was the same company that created the 35mm film format from movie film. Leica in the 1920's could have chosen a square film format then, but didn't. Leica in 2008 could have chosen a square format sensor for the S2 (as they did with the S1) but didn't.
As someone else said, square is the final preferred size too rarely.
For those of you who really want a square format, get a used Kodak 126 ( tough to get film now) or get a used Leica S1. Unfortunately, the S1 was just a scanning back, so not too good for action.
Posted by: Jack | Friday, 13 January 2012 at 02:34 PM
Interesting calculations, Eduardo.
I'm with Ed Hawco on this, as that's exactly how I approach using my 6x6 film cameras. I personally try to compose for the square but often can see that I'll need trim on an edge. Chances are that I would not be making aspect crop choices while shooting, preferring to use all the real estate and decide about such matters later.
But alas I don't think that we should expect a square sensor camera any time soon. In-camera masking, such as on the Panasonic LX5, is the most marketing-practical mass-market delivery method for this form.
But I sure would love to have a 6x6 Phase One back...
Posted by: Kenneth Tanaka | Friday, 13 January 2012 at 02:52 PM
Square sounds great, but turning a camera is some thing when hand holding takes a spit second and something you do without having to think about. When working on a tripod you're being more methodical and not really worried about speed.
As one who shot a square sensor (Hasselblad film then moved to the original Kodak DCS back) until I realized I threw away 25% of my resolution all of the time (or more since I like pano type formats), I have no interest in one. Looking back I also realize I was somewhat lazy in composition (assuming I could just decide later what I liked) and often missed something that would have helped in the final composition. (of course that's says more about me not being careful ... not really the sensors fault. but it happens.)
Posted by: Wayne Fox | Friday, 13 January 2012 at 03:04 PM
"One more benefit to having a square or round sensor, post-shot leveling. "
Square wouldn't actually help that, although round certainly would.
And I agree with Ed Hawco, that I'd rather crop in post.
Or better yet, I'd like a camera with all the options in the article as far as the viewfinder and screen go, but that still takes square pictures, and just marks the crop in the metadata.
Then the RAW software could be set to auto-crop or not giving you some options once you get into post.
Posted by: David Bostedo | Friday, 13 January 2012 at 03:11 PM
I think a square camera sensor is a very intriguing idea, but I think the ideal market for a square "sensor" is in pro-grade, high-color-gamut digital projectors. It's always annoyed me that portrait format images have to be linearly reduced by a third (relative to a 2:3 landscape portrait image) to fit the reduced vertical real estate of a landscape format projector pixel array.
When I took Art Wolfe's Art of Composition seminar, I noticed that he used a high-end Canon projector, probably Canon's Realis SX7 Mark II with an Adobe RGB color gamut:
http://www.electronista.com/articles/10/03/08/canon.outs.realis.sx7.mark.ii.projector.for.pros/
The annoying thing was that Art's portrait format images were much reduced in size and much more difficult to see at a distance. A square projection "sensor" would enable equal enlargement factors for both horizontal and veritcal format images.
At $7K a pop, this high-end projector clearly isn't cheap, but in that price range, the added cost of a larger, square pixel array would not be a deal breaker. Just a thought. Anybody have any contacts at Canon I could contact??
Posted by: Jeff Kennedy | Friday, 13 January 2012 at 03:35 PM
For Beuler,
an interesting, and correct mathematical analysis, but I wonder if you have missed another variable, that of human behaviour? If I had a square sensor, I think that every photograph would be taken with the square framing, and I would then play around with the cropping options in post-processing. Currently, I'm limited by my Nikon D200 to only 2:3 ratio, so that certainly affects the way I choose to photograph. Without a choice of framing, I rarely play too much with cropping options in Aperture. That behaviour would I think fundamentally change if all of the images came up after import in square format, and so your 90%/10% assumptions would probably not hold true _for me_.
I'm enough of an amateur to admit that my framing choices at the point of capture do not normally represent what a professional - or artist - would choose, and I'm certainly capable of chopping off feet or some interesting piece of landscape. Having the ability post-capture to fix those errors when I can see them on a 27 inch iMac screen would be welcome!
I only rarely print and frame photographs, but when I do I am not constrained by standard frame and mount sizes, as I get them made for each specific image. So standard framing proportions are not a constraint for printed output either.
Posted by: James B | Friday, 13 January 2012 at 04:53 PM
Screw the math, the real reason for shooting square is that it's the finest aspect ratio extant for portraits and always will be. Anyone who feels the long, stringy 3:2 rectangle is okay has obviously never shot square. Camera makers should be required to take Mr. Cervantes' advice and make available a version of each of their cameras with a square sensor. So, in addition to the (too) long aspect ratio of the new Nikon D4 there would also be a version, the D4sq., that would come with the bigger, square sensor. And believe me, I'm not being facetious.
And no, cropping to the square, after the fact, is not the same.
Posted by: kirk tuck | Friday, 13 January 2012 at 04:59 PM
Cameras would have to be taller to accommodate the height increase of the sensor :(
Posted by: toto | Friday, 13 January 2012 at 06:15 PM
Ed: I agree with you. "Cycling" could be distracting and time consuming. I did think about this while writing the article but I thought it was not that important at the time, and the problem I foresaw and the solution I found needed more explanation: Not having any crop at all could be risky. If you need a specific rectangle for a magazine for example, the only way to make sure it's going to work is by seeing the actual cropping. I was a heavy square shooter myself years ago and I know from then, that one can miss the imaginary rectangle often leading to badly frame or cut subjects. The solution would be to have 3 different preferences in the camera's menu. 1.-To show the entire effective square. 2.-To show the image as an actual rectangle. 3.-To show the entire square with super-imposed cropping lines in the same fashion as cropping lines over a traditional ground glass. This would be one of the sophistications necessary in a camera like this. Or bells&whistles as sometimes Americans refer to. Regards.
Posted by: Eduardo Cervantes | Friday, 13 January 2012 at 07:18 PM
I believe these cameras must already exist since I frequently see actors on television and the big screen holding their cameras horizontally and producing vertical images :).
Posted by: Bruce Stinshoff | Friday, 13 January 2012 at 08:37 PM
I believe these cameras must already exist since I frequently see actors on television and the big screen holding their cameras horizontally and producing vertical images :).
Posted by: Bruce Stinshoff
Would that be just after the car tyres screech to a halt on the dirt road? and they get out and leave the said car unlocked while heading off into the distance?
Posted by: Thingo | Saturday, 14 January 2012 at 01:04 AM
kirk tuck: And no, cropping to the square, after the fact, is not the same.
Couldn't agree more. Nor, for that matter, is building up to a square using stitching - both require you to have visualised the scene as a square, for which you can't beat having a square viewfinder.
Posted by: Tim | Saturday, 14 January 2012 at 01:28 PM
"I want to be the tripod, the light meter, the motor drive."
"The best thing about the [Leica] S2 might be that it's the only digital camera on the market with the courage to ignore "feature-anxiety," by which I mean the tendency every digital cameramaker has to anxiously lard their products up with too many features at the expense of simplicity and directness."
"The photographer could go from vertical to horizontal and to square at the soft inconspicuous touch of a dedicated button to obtain the desired framing in both the LCD or EVF."
Spot the intruder (^^;
Posted by: Bruno Masset | Saturday, 14 January 2012 at 01:51 PM
I loved shooting portraits with the square negative camera. Camera on tripod for precise framing permits one or two hands free for gestures to help direct your sitter. The problem is that sitters usually move a little leaning left or right. With a rectangular camera, re-framing is needed. With a square camera is not. Regards.
Posted by: Eduardo Cervantes | Saturday, 14 January 2012 at 03:25 PM
Hi Bruno, obviously you think that I am some sort of grinch. Please let me add: How complicated or how simple a camera is, it's up to the manufacturer and to the audience target. The most immediate use of a square sensor is to put it inside a digital back for a Hasselblad V camera. This is as simple as it gets. On the contrary, you can put a square sensor inside a Nikon D4 with all the "goodies". The Nikon D4 has 40 functional buttons, knobs and wheels all related to the picture taking process. Count them. How worse would it be 41?
Regards.
Posted by: Eduardo Cervantes | Sunday, 15 January 2012 at 10:58 AM
> Hi Bruno, obviously you think that I am some sort of grinch
Not at all.
I appreciate simplicity in cameras, but I have nothing against complicated cameras full of buttons either.
In fact, I'm of the "if I don't need a function or feature present on a camera, I just don't use it, but don't mind it being present at all, as there are probably people who find that feature useful" persuasion.
A very lenient, zen, placid and laid back school of thought, as you can see.
In this particular case, I just found a bit amusing the variety of — sometimes contrasting — opinions and perspectives one finds on TOP; this very variety, of course, is also what makes this site enjoyable :-)
I also happen to like the square format, but as far as the large manufacturers like Nikon or Canon are concerned, I'm also aware that it's quite unlikely they'd adopt such a sensor format, the reasons for which I've exposed here.
Posted by: Bruno Masset | Sunday, 15 January 2012 at 02:04 PM
HEY! This already exists!!!!
http://www.bhphotovideo.com/c/product/712755-REG/Rollei_60649_Rolleiflex_MiniDigi_AF_5_0.html
Posted by: Chad Osburn | Friday, 20 January 2012 at 10:46 AM