Why not a 3:4 aspect ratio FX *width* sensor in Nikon / Canon etc?

[geezerhood]

Since all the Nikon and Canon FX capable "full frame" lenses cover the horizontal width of the FX sensor, why not increase the size of said sensor on the vertical for a larger, higher pixel count sensor with the more appealing to many, 3:4 aspect ratio? To me that seems like the next logical step Nikon or Canon should take. It could be Canon's chance to one up Nikon if they cared to do so.

To keep the lens mount to focal plane distance the same as their current lineup, I am assuming they probably would not be able to do that in a DSLR due to the need for a "taller" mirror and thus a deeper mirror box, but maybe it might fit? If it would that would be wonderful.

What else is preventing such a move in a DSLR body that works with all current Nikon or Canon lenses?

If that feat is impossible then a mirrorless body with an FX width, 3:4 aspect taller sensor with the same pixel density as the sensor in the D810, to fit the current crop of Nikon and / or Canon lenses would be an instant hit would it not?  

If priced between the D810 and the 645z, I would get one for sure.

What would the pixel count be for such a sensor? (It's too early for me for maths)

[Telecaster]

With a 36x27mm sensor you'd likely see increased corner light falloff with some lenses…not necessarily a deal-killer but something to be considered. Personally I'd like to have a 32x32mm sensor. You'd get good coverage of the whole thing with (non-baffled) existing 35mm format lenses, plus 32x24mm (or 32x21.3mm, etc.) horizontal or vertical frames without needing to spin the camera 90°. IMO all of this would work far better with a mirrorless system than an SLR.

-Dave-

[mcbroomf]

A lens really covers a diameter = the diagonal of the 24x36 frame which is 43 1/4 mm.  The diagonal of a 36x27 frame is 45mm.  Some lenses may work, others not so much.  To keep the diameter the same, a 3:4 frame would have to be 26 x 34.6mm.

[Torbjörn Tapani]

I would like a 3:4 sensor or square even. Or just a medium format mirrorless. It would be expensive (sensor is the biggest cost in a FF camera) but you could pick a crop freely if your lens can handle it. Some lenses have over sized image circles(the Zeiss 135 as we saw today) Others would work fine on one end of a zoom range. It would be no different than using DX lenses on a FX camera. I have a DX fisheye for example that I got specifically to use on FX.

[BJL]

As some have said, some lenses would have poor corner performance. Also, some have internal anti-flare baffles that would severely vignette.  But that is not fatal: at worst, some images would have to be cropped back to about 36x24mm, or some other shape like about 34x26mm, while with others lenses you could benefit from the larger frame size.

But this is probably only viable with mirrorless systems: in an SLR, the extra frame height would reuire a deeper mirrorbox that is probably incompatable with existing lens mounts and lenses, along with a higher viewfinder, and at that point, most people aspiring to shapes like 4x3 or popular print shapes like 10x8 or 7x5 or 11x8.5 or 20x16 or A4 or A3 etc. will prefer to continue with what they have been doing for almost a century with the 36x24mm format: crop a bit at the sides.

So, expect it from Sony before Canon or Nikon.

[dwswager]

There are a host of technical reasons why Nikon doesn't deviate from 36x24 (3x2) format.  Mirror box and Shutter curtain issues being primary ones as well as lens issues.

Look, we already have a 36MP 3x2 sensor.  At square crop, the D810 gives a 24.12MP (4912x4912) image.  While square is useful for some disciplines, and seems to work well in news and web publication, the unequal proportion is more useful while requiring less size, weight cost and throughput capacity.  It does require a conscious decision be made at the time of capture.  I understand the desire to shoot one image and crop it landscape or portrait all from that single frame.

Assuming we get a 3x2 full frame sensor with the 3.9 micron pixel pitch of the 24MP D7100 DX senosr, then we would more than 6000 pixels square inside that 3x2 frame!

[BernardLanguillier]

Besides, there are many cases where a 2 frames handheld stitch is an option if cropping a 36mp to a square 24mp image contains too little resolution.

Cheers,
Bernard

[allegretto]

Yep, I can see them taking a chance on a shutter redesign, sensor redesign, body redesign, mirror redesign (and with the axis switch it has to move even faster and stop from that higher velocity without too much slap), chubbier body just because it might please a few folks... srsly...?

I don't see any major actually doing this any way but an electronic crop in camera which you can do in Post anyway...

[Bart_van_der_Wolf]

Since all the Nikon and Canon FX capable "full frame" lenses cover the horizontal width of the FX sensor, why not increase the size of said sensor on the vertical for a larger, higher pixel count sensor with the more appealing to many, 3:4 aspect ratio

Hi,

Besides the technical reasons already mentioned by others, I wonder about the appeal of a 4:3 ratio.

If we are to follow the studies from Leonardo DaVinci and contemporaries, a Golden Section (~1.618:1, or ~3.2:2) ratio is probably more appealing (if we want to restrict ourselves to something that rigid). However, personally I tend to let the subject (e.g. landscapes, architecture) dictate the best ratio, unless I deliberately need to frame a subject in a predefined space. Even with commonly available standard frame sizes, a matte will allow to fit any aspect ratio in that space.

At shooting time, a bit of change in the perspective viewpoint may allow the composition to fit better in a fixed ratio frame, but then we're back at the initial question, which ratio, or why 4:3?

Cheers,
Bart

[dwswager]

At shooting time, a bit of change in the perspective viewpoint may allow the composition to fit better in a fixed ratio frame, but then we're back at the initial question, which ratio, or why 4:3?

Cheers,
Bart

Overall, well said.  Personally, if it wasn't 3x2 as 35mm film was, I would opt for square.  Assuming a high enough pixel density, square would allow you to crop later and leave more options open.  Of course, the size and cost of square would be a limiting factor to the appeal.  Pretty much why 35mm was so much more popluar than 2 1/4 inch back in film days.  BTW, my wedding was shot on 2 1/4" Hassy's and I understand the appeal of leaving your options open.

On the flip side of that argument, I think 3:2 makes you consider and explore the photo opportunities as they present themselves.   I know lots of times I think oh, this is a natural vertical and shoot that first, and then pop off a horizontal only later to like the horizontal framing better.

[Telecaster]

In the real world I'm happy with any camera system, whether it uses a 3:2 or 4:3 sensor, that allows in-finder cropping to alternate aspect ratios. This is one reason why I've gravitated to mirrorless systems. I use my A7r in 16:9 a lot…and I wish it offered 4:3 & square as well. m43 cameras offer all these ratios, and depending on the subject & situation I use 'em.

-Dave-

[geezerhood]

Thanks all for your comments.  I figured that if a lens could cover the horizontal width then it would also be able to cover the same distance vertically, but I didn't take into account design factors such as baffles as mentioned by BJL that would be cutting off bits of the image circle at the top and bottom

So it looks like it wouldn't work very well if at all with the current lens designs out there.

Too bad.

[BJL]

Hi,

Besides the technical reasons already mentioned by others, I wonder about the appeal of a 4:3

If we are to follow the studies from Leonardo DaVinci and contemporaries, a Golden Section (~1.618:1, or ~3.2:2) ratio is probably more appealing ...

This myth keeps coming up; it might be relevant to architectural shapes (which is where that ideal arose in classical architcture) but as soon as there is an image to be placed in a rectangle, the facts show a quite differemt pattern of preferences. I have looked at the actual shape choices of paintings, drawings, and photography over many centuries, through examining the published dimensions in art books and show catalogs, and measuring reproductions in such books when the specs are not given, and have examined such "rectangular art works" from a wide range of eras, medieval to 20th century. The consistent result is that the dominant range of shapes is from 5:4 to 3:2, (1.25 to 1.5) with the mode about 4:3 (1.33) or a bit higher, but closer to 4:3 than to 3:2. If anything, there is a gap around the Golden Ratio (about 1.62), with a jump from 3:2 to highly panoramic shapes like 2:1 and beyond.

Another indication is the array of shapes for canvasses and drawing paper in art supply storees, and of course the shapes of photographic and inkjet printing paper: the same favorites of 5:4, 4:3, 7:5 and 3:2 dominate, and for canvasses and drawing pads, 4:3 seems to be the most popular shape, beating 3:2 into second place. Canvas shape options are more diverse than photographic ones, with the main extra shapes being square, 6:5, and 2:1.

Once you think about verticals as well as horizontals, the nonsense about the ideal shape being beyond 3:2 shpuld become clear, but even looking only at horizontals, my survey shows that dominance of the range 5:4 to 3:2, with a mode between 4:3 and 7:5.

If the Golden Ratio were so appealing for "rectangular artistic images", one would expect simple nearby shapes like 8:5 to be offered, but I have never seen canvasses, drawing pads or photogrphic printing paper offered in that shape, or in any shape between 3:2 and 2:1, other than the outlier of 17"x11", derived from doubling US letter size of 11"x8.5" and probably originally for two-up printing for folding to 11x8.5.


P. S. Hopefully we will not get side-tracked by the different shape choice for _moving_ pictures.

[Telecaster]

It's good to keep in mind too that the 36x24mm frame, at least as chosen by Barnack for the Leica, comes from rotating a ~1.33:1 motion picture frame by 90° (along with the film itself) and then butting a pair of 'em together.    This was primarily a practical choice.

Very few photos I take with 3:2 sensors or film frames end up using that ratio after processing. When I scanned 35mm film I liked 7:4 for many horizontals and 4:3 or 5:4 for most verticals. Once 16:9 TVs came on the scene I switched from 7:4…the two are very close anyway. I should add that these ratios are starting points…I enjoy making photos work within their bounds but I'll tweak if/when necessary.

-Dave-

[allegretto]

actually, most is either Landscape of various ratios or Portrait of various ratios

Most of us recall the story about horses asses determined the size of Space Shuttle boosters. But that does''t mean it's true

Many artists worked in Portrait configurations too.

The salient point is that a camera would have to be near-totally redesigned to "go vertical" and I just don't see them doing that without a demand to just see what happens if they Market that.

This myth keeps coming up; it might be relevant to architectural shapes (which is where that ideal arose in classical architcture) but as soon as there is an image to be placed in a rectangle, the facts show a quite differemt pattern of preferences. I have looked at the actual shape choices of paintings, drawings, and photography over many centuries, through examining the published dimensions in art books and show catalogs, and measuring reproductions in such books when the specs are not given, and have examined such "rectangular art works" from a wide range of eras, medieval to 20th century. The consistent result is that the dominant range of shapes is from 5:4 to 3:2, (1.25 to 1.5) with the mode about 4:3 (1.33) or a bit higher, but closer to 4:3 than to 3:2. If anything, there is a gap around the Golden Ratio (about 1.62), with a jump from 3:2 to highly panoramic shapes like 2:1 and beyond.

Another indication is the array of shapes for canvasses and drawing paper in art supply storees, and of course the shapes of photographic and inkjet printing paper: the same favorites of 5:4, 4:3, 7:5 and 3:2 dominate, and for canvasses and drawing pads, 4:3 seems to be the most popular shape, beating 3:2 into second place. Canvas shape options are more diverse than photographic ones, with the main extra shapes being square, 6:5, and 2:1.

Once you think about verticals as well as horizontals, the nonsense about the ideal shape being beyond 3:2 shpuld become clear, but even looking only at horizontals, my survey shows that dominance of the range 5:4 to 3:2, with a mode between 4:3 and 7:5.

If the Golden Ratio were so appealing for "rectangular artistic images", one would expect simple nearby shapes like 8:5 to be offered, but I have never seen canvasses, drawing pads or photogrphic printing paper offered in that shape, or in any shape between 3:2 and 2:1, other than the outlier of 17"x11", derived from doubling US letter size of 11"x8.5" and probably originally for two-up printing for folding to 11x8.5.


P. S. Hopefully we will not get side-tracked by the different shape choice for _moving_ pictures.

[Bart_van_der_Wolf]

If the Golden Ratio were so appealing for "rectangular artistic images", one would expect simple nearby shapes like 8:5 to be offered, but I have never seen canvasses, drawing pads or photogrphic printing paper offered in that shape, or in any shape between 3:2 and 2:1, other than the outlier of 17"x11", derived from doubling US letter size of 11"x8.5" and probably originally for two-up printing for folding to 11x8.5.

Hi,

The usual problem with such analyses is the disconnect between correlation and causation. When there are no Golden Ratio image formats offered (e.g. paper or canvas size), then they will not appear in the statistics. That says nothing about their visual appeal. One would first have to establish if an output size is appealing enough to produce such an option, and it's apparently not that commonly requested.

The difficulty with that is that the golden section is also about relative subject dimensions and placement, and equally applies to a square output as it does to a rectangular output. Some subjects, e.g. landscapes may be represented in a more panoramic output size to stress the wide vista and rolling hills, but elements within that composition (e.g. the position of the horizon/vanishing points) may be preferred to follow the golden section principle if a pleasing rendering is desired, or deliberately not (e.g. dead center, or close to an edge, or even partially cropped) if tension and an uneasy feeling is required. There is also a difference between placement at the exact section lines, or subdivisions of regions in which to place.

Here is an interesting paper by Prof. McManus that covers two angles, namely the individual preferences for for simple figures, and also that of composition in a frame. Another paper explores the findings that while the aesthetic prefererence  of a population may be centered at the modal point around a golden section rectangular shape, there are also strong individual preferences for other shapes.

So things may be a bit more complicated than some might wish, but there is something special about the Golden section (and Fibonacci ratios) because it occcurs so often in nature. It has also been used for book design proportions ("Many books produced between 1550 and 1770 show these proportions exactly, to within half a millimeter").

Cheers,
Bart

[Bart_van_der_Wolf]

As fine art students the Golden Mean was brought to our attention but we were then encouraged to ignore it. However when we were asked to bisect a given line or place an element within a given frame it was interesting how often our selection approximated the rule.

Indeed, one can only deviate from something that is a baseline/'rule' to begin with (not that it is a real rule, it's more a recipe that can be deviated from if one knows what one is doing, seasoned to taste). Understanding why it works so well then allows to deliberately do something different. And frankly, when I look at my compositions that are intended to be 'pleasing', rather than jolting an emotion of unease, it's usually pretty close to Golden section divisions and placement of the key elements.

However, that is within the frame that fits the subject. Some subjects look better in square frames, others in stretched rectangular frames (or other shapes like a diamond, or combinations like a tryptish).

Cheers,
Bart

P.S. Some more opinions and research.

[kers]

Nikon in a d810 lets you choose between  36 x24mm and  30x24mm ( 5:4) ,
then they have a 30x20mm and a DX mode 24 x16mm - so enough to choose from...

[geezerhood]

My thoughts on increasing the size of the Sony sensor used in the Nikon D800 series, on the vertical dimension were not just to change the aspect ratio. I saw it as a possible way to get an increase in the number of pixels you end up with in the image and to improve the FX frame to a more appealing ratio at the same time. I don't want to crop if I don't have to as I feel that the 36 mega-pixel file is BARELY enough to make me happy with my large prints and I don't enjoy stitching.

They are both equally important to me personally. I still wonder how many of the Nikon lenses would actually cover a 4:3 sensor that maintained the full width of the FX.

If I did the math right then it looks like the pixel increase would be to almost 41 mega-pixels with a 4:3 ratio, probably not enough to make it attractive to many.
If the lenses for the Nikon or the Sony could cover 1:1 at FX width, then it would be over 54 mega-pixels which would definitely draw some interest considering you would not have to buy any lenses to go with the body and you could choose your aspect ratio and orientation without sacrificing image quality if that matters to you. But it looks like none of the current lenses would cover 1:1, sadly.  The most likely candidate for a square format 54 mega-pixel body is probably Sony with a new series of lenses to match it. I can't imagine Nikon ever doing something outside the FX format. If it was priced similar to their current 36mp body i would buy one.

As for golden triangle ratios and such, and what appeals to me, I also do work with oils, watercolors, pastels, intaglio, block prints, pen and ink etc and can create a work surface in virtually any shape or size I wish for all of those mediums, building frames to match. I rarely (probably NEVER if I were to look closely) choose an aspect ratio close to that of the FX sensor as it does not appeal to me, especially in a vertical orientation. It just "feels" unbalanced. I much prefer 4:3 or 5:4 and tend to stick to those most of the time or I will go substantially shorter on the smaller dimension of the FX ratio, if i want something wider or skinnier looking.

[Theodoros]

A lens really covers a diameter = the diagonal of the 24x36 frame which is 43 1/4 mm.  The diagonal of a 36x27 frame is 45mm.  Some lenses may work, others not so much.  To keep the diameter the same, a 3:4 frame would have to be 26 x 34.6mm.

Very well said... Which leads us to conclude that it's best (financially) for the maker to only include a 32mm x 24mm crop ratio in the menu, since resolution decrease will be insignificant to consider... Otherwise, cost increase to make a 26x34.6 camera is much more than one should imagine for the following reasons:
1. A different mirror box and prism is required
2. A different shutter mechanism is required
3. Different mount to image area distance is required, which would mean new lens line
4. Flash sync speed will have to be for a slower speed
5. There would be more sensor cropped out in video mode
6. All (new) P/C lenses would perform worst than they now do.
7. Shutter life would decrease