Here's an idea - don't produce an FF camera

It should never be forgotten that there is nothing sacred about the 35mm format. It was famously made popular in still photography by Leica in the 1920s, being adopted from a film format that we invented at the end of the 19th century. Leica used it as a compact, easily portable format which was eagerly taken up in the latter half of the 20th century as an ideal compromise between quality and portability. There are still some fundamentals of optics which means that, even in the digital age, it's probably at a favoured position allowing a good balance of portability, size, image quality and artistic control. There is also a huge legacy investment of 35mm compatible lenses out there representing an accumulated expenditure of heaven knows how many billions of dollars.

We could view this legacy as tens of millions of image circles, all roughly 43.3mm in diameter. That's about 1,470 sq mm, of which the 24 x 36 format uses 864, or about 59%. I suspect that there's probably even more imaging real-estate available than that - especially with vignette and other corrections, albeit that this will be lens-model dependent. A 10% increase in image circle (2.3mm edge margin) would yield a 21% increase in area.

Now we have a near fully digital era, EVF is now a viable option and the moving mirror can be omitted, an innovative camera manufacturer could break free of the constraints of the legacy 3:2 frame format. Some of these formats might present a challenge in accommodating an SLT mirror if that is the type of camera adopted, but not with a mirrorless design. The shutter would also have to have a larger vertical coverage. Note that SSS is not affected in that the required degree of movement is dependent on the lens focal length and shutter speed, not the sensor size so the same image circle "tolerance" is required as for FF.

1) go the conservative route of a 24 x 36mm frame. At 4.75 micron (A580) pitch, this yields 38MP.

2) adopt a square format sensor with a diagonal of 43.3mm. The maximum frame would measure 30.6 x 30.6 and yield 41.5MP at 4.75 micron pitch. It would also support portrait and landscape modes (at other crop factors, such as 3:2, 4:3 and 16:9) without requiring rotation of the camera, albeit at reduced MP count. The 3:2 format, for instance, would measure 20.4 x 30.6 and yield 27.7MO.

3) adopt a square format sensor of 43.3 x 43.3mm. This would support square 30.6 x 30.6mm (41.5 MP) plus portrait and landscape format 3:2 formats at 24 x 36mm (38MP) plus, of course, an number of other aspect ratios (4:3, 16:9 etc.) and crop factors provided that the picture diagonal did not exceed 43.3 mm. If it turned out vignetting was acceptable (or could be compensated for) it might even be possible to increase frame sizes beyond 43.1 x 43.3mm

It would be interesting to know if (3) is viable. I would imagine that a camera which allowed complete fluid control of aspect ratios and crop factors without having to rotate the camera would have lots of advantages - not least in the studio. Of course the large square sensor option would be more expensive, and it would require square aspect backscreens and EVFs (with electronic cropping), so it wouldn't be cheap. However, it would to start make use of that multi-billion legacy lens stock...

Your first option: "1) go the conservative route of a 24 x 36mm frame. At 4.75 micron (A580) pitch, this yields 38MP." That is what is called a Full Frame sensor. The same size currently used in all full frame cameras like the A900, Nikon D3, Canon 5D and Canon 1Ds.

Your 2nd and 3rd options of adopting a square format sensor can be made using the same mount and the same lenses but probably only up to 30mm x 30mm. Since right now the lens is only utilizing part of the lens coverage area (24x36mm). There is no problem making the sensor square if they want with the same mount. Just more expensive than a full frame sensor since is going to use even more silicon.

I'm full aware option 1 is FF - which is why I called it all the conservative option. As far as the rest goes, all the frame sizes I've calculated fit within the image circle which has a diameter of 43.3mm (approx).

Option (2) is the largest square format (30.6 x 30.6) which fits within that size.

Option (3) is, of course, oversized but the point of that was not that it would allow 43.3 x 43.3mm images to be taken, but that it would allow a multitude of crops to be taken such as 24 x 36mm in landscape and portrait mode without requiring re-orientation of the camera. Indeed the whole point that this was just two of very many formats 16:9, 4:2, square and so on that could be produced in either vertical or horizontal orientation without rotating the camera provided that the diameter didn't exceed 43.3mm. Indeed the whole idea of the "over-square" format is to allow the maximum flexibility of framing within the available image circle.

... given that electronic cropping presets would be best composed upon when clearly visible in the viewfinder. Of course, a distinct masking LCD overlay on an OVF would provide that same utility.

A mirror supporting a 30*30mm or even 43*43mm format OVF would be big, though, and will not fit into the flange distance between lens and sensor if you intend to keep on using legacy lenses. So SLT and/or mirrorless EVF might be the only way ahead by that mechanical limitation.

Intriguing in this square sensor concept is that framing in landscape or portrait mode electronically might help in getting rid of a lot of brackets, riggs, gimbals, and all sorts of mechanical stuff that gets in the way when reframing by 90 degrees with current fixed aspect ratio cameras.
--
Ralf
http://RalfRalph.smugmug.com/

SteveGJ said:

It should never be forgotten that there is nothing sacred about the 35mm format.

Click to expand...

Nothing "sacred" but then again, it's the format used by some 100+ million AF lenses sold by Canon, Nikon and Minolta, and the format for which the broadest systems are now available. The "legacy" as you say. (The fact that it's not sacred is illustrated by competing systems).

SteveGJ said:

Now we have a near fully digital era, EVF is now a viable option and the moving mirror can be omitted,

Click to expand...

That's big - even going to a 30x30 square would have resulted in a taller mirror flapping back & forth.

SteveGJ said:

1) go the conservative route of a 24 x 36mm frame. At 4.75 micron (A580) pitch, this yields 38MP.

Click to expand...

Probably the preferred route since some lenses utilize baffles to reduce internal flare, and these baffles assume a 24x36mm sensor/film. In other word, the optics may be designed for a circle with a 43.3mm diameter, but the image that hits the sensor is not circular. Also, many lens shades will result in additional vignetting with a taller sensor. You could modify production of new lenses & shades to get around this, but then you're potentially sacrificing image quality (contrast, anyway).

SteveGJ said:

2) adopt a square format sensor with a diagonal of 43.3mm. The maximum frame would measure 30.6 x 30.6 and yield 41.5MP at 4.75 micron pitch. It would also support portrait and landscape modes (at other crop factors, such as 3:2, 4:3 and 16:9) without requiring rotation of the camera, albeit at reduced MP count. The 3:2 format, for instance, would measure 20.4 x 30.6 and yield 27.7MO.

Click to expand...

Better than APS-C and potentially interesting, but unless you crop "squarer" than 4:3 frequently, 24x36 still gives you more real estate.

SteveGJ said:

3) adopt a square format sensor of 43.3 x 43.3mm. This would support square 30.6 x 30.6mm (41.5 MP) plus portrait and landscape format 3:2 formats at 24 x 36mm (38MP) plus, of course, an number of other aspect ratios (4:3, 16:9 etc.) and crop factors provided that the picture diagonal did not exceed 43.3 mm. If it turned out vignetting was acceptable (or could be compensated for) it might even be possible to increase frame sizes beyond 43.1 x 43.3mm

Click to expand...

It's an interesting idea, but medium format shows that the cost would be prohibitive*, and you have the lens issue I mentioned above.

• Then again with the Pentax 645 at $10K and the Sigma SD1 at $7K, there could be room for a very expensive camera that utilizes legacy SLR lenses.

• Dennis

--
Gallery at http://kingofthebeasts.smugmug.com

Although it will be more expensive than a regular FF sensor it will be a great option if Sony decide to make a 30mm x 30mm square sensor. It may be even able to compete favorably with some of the established medium format cameras like the Pentax 645. Specially since it can use the same 35mm SLR mount and lenses. And you can be sure Nikon is going to want to make a camera with it also.

Assuming frequent crops from square (i.e. your main purpose is maximizing IQ for a number of aspect ratio and avoiding rotating the camera) you have the same issue with the EVF that you do with the sensor: namely that however big you make the EVF, unless you're shooting square, you're always cropping. So to get a 2:3 VF image as big as what's in the A77 now, you already need an EVF that's 50% bigger. And that's still not as big as the OVFs in current FF cameras.
--
Gallery at http://kingofthebeasts.smugmug.com

You can use the spare space on the EVF for exposure and navigation information that would otherwise clutter up the image space. Of course the information would have to be moved according to the actual aspect ratio and orientation, but it should be possible in principle. A square format EVF of the same sort of quality as the A77 would be around the 3MP mark if it was to allow for 3:2 aspect ration with the same resolution.

The 35mm lenses that I have show no signs of any baffles impinging on the image circle. It's fairly easy to test; wedge the aperture open and focus a bright image onto a sheet of paper. Rotate the lens and look for any differences in brightness. I think I'd expect to see such effects looking through the lens as well.

Maybe there are lenses out their with baffles, but I can't see any sign of it in my collection (admittedly all but one are Sigma lenses). I have an (old) dismantled Sigma 70-300mm zoom as my A700 stripped the gears within a minute. It certainly has not baffles in it.

SteveGJ said:

The 35mm lenses that I have show no signs of any baffles impinging on the image circle. It's fairly easy to test; wedge the aperture open and focus a bright image onto a sheet of paper. Rotate the lens and look for any differences in brightness. I think I'd expect to see such effects looking through the lens as well.

Maybe there are lenses out their with baffles, but I can't see any sign of it in my collection (admittedly all but one are Sigma lenses). I have an (old) dismantled Sigma 70-300mm zoom as my A700 stripped the gears within a minute. It certainly has not baffles in it.

Click to expand...

A couple of examples -- just scroll down to the pictures showing the rear end of the lens ...

http://www.kurtmunger.com/tamron_70_300mm_f_4_5_6id290.html
http://www.kurtmunger.com/sony_70_300mm_f_4_5_5_6_gid224.html

I've seen many lenses having this sort of rectangular baffle just aft of the rear element.

Cheers,
--
Greg

It it's that obvious I can confidently claim none of my lenses have that feature, including my sole Tamron (18-250 Di II); not that it would be much use on large sensors of course, but it could do an APS-C equivalent.

SteveGJ said:

It should never be forgotten that there is nothing sacred about the 35mm format. It was famously made popular in still photography by Leica in the 1920s, being adopted from a film format that we invented at the end of the 19th century. Leica used it as a compact, easily portable format which was eagerly taken up in the latter half of the 20th century as an ideal compromise between quality and portability. There are still some fundamentals of optics which means that, even in the digital age, it's probably at a favoured position allowing a good balance of portability, size, image quality and artistic control. There is also a huge legacy investment of 35mm compatible lenses out there representing an accumulated expenditure of heaven knows how many billions of dollars.

Click to expand...

Undoubtedly true.

SteveGJ said:

We could view this legacy as tens of millions of image circles, all roughly 43.3mm in diameter. That's about 1,470 sq mm, of which the 24 x 36 format uses 864, or about 59%. I suspect that there's probably even more imaging real-estate available than that - especially with vignette and other corrections, albeit that this will be lens-model dependent. A 10% increase in image circle (2.3mm edge margin) would yield a 21% increase in area.

Now we have a near fully digital era, EVF is now a viable option and the moving mirror can be omitted, an innovative camera manufacturer could break free of the constraints of the legacy 3:2 frame format. Some of these formats might present a challenge in accommodating an SLT mirror if that is the type of camera adopted, but not with a mirrorless design. The shutter would also have to have a larger vertical coverage. Note that SSS is not affected in that the required degree of movement is dependent on the lens focal length and shutter speed, not the sensor size so the same image circle "tolerance" is required as for FF.

1) go the conservative route of a 24 x 36mm frame. At 4.75 micron (A580) pitch, this yields 38MP.

Click to expand...

I like that as you know.

SteveGJ said:

2) adopt a square format sensor with a diagonal of 43.3mm. The maximum frame would measure 30.6 x 30.6 and yield 41.5MP at 4.75 micron pitch. It would also support portrait and landscape modes (at other crop factors, such as 3:2, 4:3 and 16:9) without requiring rotation of the camera, albeit at reduced MP count. The 3:2 format, for instance, would measure 20.4 x 30.6 and yield 27.7MO.

Click to expand...

You have crop factors confused with aspect ratios. Crop factors are related to the the effective focal length of the lens translated from one sensor size to another. - for instance a 200 mm full frame lens becomes equivlent to a 300 mm lens at a 1.5 crop factor for an APS sensor. Another way or looking at that is the crop factor relates to the relative angle of view for a scene.

SteveGJ said:

3) adopt a square format sensor of 43.3 x 43.3mm. This would support square 30.6 x 30.6mm (41.5 MP) plus portrait and landscape format 3:2 formats at 24 x 36mm (38MP) plus, of course, an number of other aspect ratios (4:3, 16:9 etc.) and crop factors provided that the picture diagonal did not exceed 43.3 mm. If it turned out vignetting was acceptable (or could be compensated for) it might even be possible to increase frame sizes beyond 43.1 x 43.3mm

It would be interesting to know if (3) is viable. I would imagine that a camera which allowed complete fluid control of aspect ratios and crop factors without having to rotate the camera would have lots of advantages - not least in the studio. Of course the large square sensor option would be more expensive, and it would require square aspect backscreens and EVFs (with electronic cropping), so it wouldn't be cheap. However, it would to start make use of that multi-billion legacy lens stock...

Click to expand...

Not really a bad idea overall. However it would have the most appeal to pros (which are the minority) and much less so to the majority of amatures. One reason is because of the shape of our monitors and HDTV screens. Old habits die hard.

I don't think a lot of appeal among the majority of the purchasers would come from lowering effective megapixels from 38 to 27 for our 3:2 images, as you already know how crazy our community is about that. The cost would be very high - for one reason being just because manufacturers can get it. The format would still get stiff competition from medium format cameras - which would also up their number of MP and in those the narrow depth of field in commercial and professional work would swing the balance away from a square sensor 43 mm diagonal or 30mm sensor. But I see the intelligence in what you propose.

Some people like square format in the medium format class. I don't know why, but since I always print in USA traditional sizes 8x10, 11x14, and 16x20 even my medium format cameras (Pentax 645 and Fuji GW690-III) follow that rectangular format.

It should never be forgotten that there is nothing sacred about the APS-C format. It was famously made popular in still photography by Canon in the early 21st century, being adopted from a film format that was invented at the end of the 20th century. It was introduced as a compromise between image quality and affordability. There is now a huge legacy investment of DT lenses out there representing an accumulated expenditure of heaven knows how many millions of dollars.

In theory I support your idea about square sensors, but in practice, how many picture frames are square? To get the most out of the sensor, it should have the same relative dimensions WxH as most of the final images. You can just as easily crop a square from a rectangle as vice versa.