Square sensors impossible?

[J A C S]

I should have stated immediately that it should be oversized such as 18x18 for M43, 24x24 for APS and 36x36 for FF. Doesn't make sense otherwise. The corners of the sensor are wasted whereas you gain now wasted parts of the image circle. A clear win to me. I'm aware that FF is pushing the border of affordable sensor size so 36x36 might not be viable yet except as a medium format alternative with similar pricing but MF is coming down too. Another qualification is that it's not for DSLR as there is no space available for the bigger mirror. Nevertheless there's always been mirrorless where it's a logical solution.

It will cover some of the contacts...

 

[Christof21]

Hello,

More precisely: what makes sense is a multi ratio sensor...

It actually makes no sense. The sensor is a hard physical object with millions of electronic connections. How could anyone ever alter its size?

For instance a FF sensor with aps-c lens can have many advantages, you get more light as soon as you select a ratio different from 3:2.

This isn't a variable sensor

I haven't said this

- it is selecting different areas from the same sensor.

Yes

But that sensor has to be as big as the biggest frame you want to use from it, so this in no different from providing an over-sized sensor.

This is what I meant, I even gave the size needed for the sensor.

Not so over-sized according to the math, an increase of 27.5% of the height is enough.

 

[rurikw]

I should have stated immediately that it should be oversized such as 18x18 for M43, 24x24 for APS and 36x36 for FF. Doesn't make sense otherwise. The corners of the sensor are wasted whereas you gain now wasted parts of the image circle. A clear win to me. I'm aware that FF is pushing the border of affordable sensor size so 36x36 might not be viable yet except as a medium format alternative with similar pricing but MF is coming down too. Another qualification is that it's not for DSLR as there is no space available for the bigger mirror. Nevertheless there's always been mirrorless where it's a logical solution. In the beginning of digital, sensors were probably much more expensive and made up a larger part of the price of the camera so it made sense to stick to film logic and statistics of format preferences thus saving crucial mm^2s of sensor area but that hasn't been the case for a long time now.

For about 50 years a very (? the most) popular style of camera was an SLR. This used the frame size of 36 x 24mm that became popular 100 years ago and still is.

I'm sure it is but digital should give us more freedom.

To make a camera with film or sensor 36mm high would mean increasing the height of every one of those cameras. For (D)SLRs it would also mean that the body would have to be 12mm deeper to accommodate the larger mirror; and that would mean that existing lenses would be needed with a longer flange distance. In other words, not an economic proposition.

As I said, not so much for DSLRs.

Moreover, to utilise the full area of the 36 x 36mm frame the image circle is 51mm* compared to 43mm

Yes but that's irrelevant. Did you notice I talked about an oversized sensor?

so either:

(a) a whole new series of bigger, heavier (and more expensive) lenses would be needed or

not needed, see above

(b) much of the area of the sensor would be wasted. In fact, the square to fit inside 43mm is just under 26mm so all you get is a fractional increase over the 24mm currently available.

your calculation is off, it's more than 30x30 so an increase of over 50%

* Note that many medium format digital cameras require 55mm image circle so the new camera would cost close to the cost of medium format digital, camera.

We have been through this before. I don't know why we keep talking past each other. Last time I sent you a chart for M43 with a 18x18 sensor. Now I show you one for FF. Same thing scaled up (+ the little difference between 4:3 and 3:2).

As I said going beyond 36x24 might be expensive at the moment so the concept is probable more viable for APS. You can easily scale down the above chart, it works the same way. There is a small error in the diameter (44 vs 43.27) but that dos not affect the overall picture.

 

[Tom_N]

Another qualification is that it's not for DSLR as there is no space available for the bigger mirror.

If you are designing a new camera to take a larger sensor, you MAKE room for the bigger viewfinder and the bigger mirror. It's not as if the manufacturer is going to offer to put a larger sensor into the DSLR or MILC that you have now.

First it has to make sense to design a camera to take a larger sensor / different AR.

 

[Tom_N]

Largest 3:2 rectangle that will fit inside a FF image circle: 36mm x 24mm

Largest 1:1 square that will fit inside a FF image circle: 29.8mm x 29.8mm

Largest 3:2 rectangle that will fit inside that square: 29.8mm x 19.9mm

(Same idea applies for APS-C or M4/3, just with different numbers.)

Now you could go for a 36mm x 36mm sensor and use only a 43mm wide part of it for any given shot, but then you're talking a 50% larger sensor area and reduced yield. Also, in any scenario where the capture area is taller than 24mm, there is a possibility that some lenses that have baffles will exhibit vignetting at the top and bottom of the frame.

This means that to cover FF image circle for ratios from 3:2 to 1:1, a 36x30 sensor would be optimal.

How does this match up with what people say about wanting square sensors so that they can take portrait photos without losing effective resolution or rotating the camera?

With a 36x30 sensor, your non-rotated 3:2 portrait exposures would cover at most 20mm (width) by 30mm (height) of the sensor. Slightly better than the 16x24mm that you could get today, but still, not the 24x36mm that most square sensor advocates hope for.

 

[Mark B.]

Hello,

Sorry if this question was asked/answered previously.

Indeed it has. Lots and lots, lol.

To a newbie it seems a square sensor would make a lot of sense in practical terms, including a slight increase in actual area captured. I understand that a lot of other things like TV's and displays have rectangular dimensions. However, a square sensor would offer some advantages it seems?

Like cropping every single image? Surprisingly not many people want to do that.

I hear a long time ago sensors/photographic plates were square and people were ok with that.

That really was a very long time ago, when view cameras were the norm and it would have been awkward to rotate the camera.

I understand that old ways of producing silicone wafers would have to be changed, but isn't that what innovation all about? Is such a change really cost prohibitive? It would be a dream for me to not have to flip the camera back and forth in different orientations, just shoot and later crop. I supposed others prefer different?

Vignette would decrease too, you get more pixels.

Thanks for the education.

In reality it would be a very niche market in today's world. Very low demand for a square sensor would make the production and retail cost very high. Certainly there will be responses from folks here saying they would like it, but this is a tiny fraction of the market and not a good representation.

 

[Tom_N]

To make a camera with film or sensor 36mm high would mean increasing the height of every one of those cameras. For (D)SLRs it would also mean that the body would have to be 12mm deeper to accommodate the larger mirror; and that would mean that existing lenses would be needed with a longer flange distance. In other words, not an economic proposition.

Moreover, to utilise the full area of the 36 x 36mm frame the image circle is 51mm* compared to 43mm so either:

(a) a whole new series of bigger, heavier (and more expensive) lenses would be needed or

(b) much of the area of the sensor would be wasted. In fact, the square to fit inside 43mm is just under 26mm so all you get is a fractional increase over the 24mm currently available.

Closer to 30mm than 26mm, but otherwise, good points.

Then consider that if you are willing to rotate the camera to match the photo orientation, you can already use a lot of the FF sensor area for other common aspect ratios:

• 7:5 – 33.6 x 24mm
• 4:3 – 32.0 x 24mm
• 5:4 – 30.0 x 24mm

No camera modification needed, although perhaps optional grid lines (for composition) would be nice.

* Note that many medium format digital cameras require 55mm image circle so the new camera would cost close to the cost of medium format digital, camera.

 

[rurikw]

Another qualification is that it's not for DSLR as there is no space available for the bigger mirror.

If you are designing a new camera to take a larger sensor, you MAKE room for the bigger viewfinder

Yes of course

and the bigger mirror.

Not a good idea. That would increase the flange distance and require a whole new system.

It's not as if the manufacturer is going to offer to put a larger sensor into the DSLR or MILC that you have now.

Basically it is. Size of the sensor, shutter and displays + whatever changes to the electronics and proportions of the camera are needed but the lenses would be the same, only you would get more resolution and AoV out of them.

First it has to make sense to design a camera to take a larger sensor / different AR.

To me it makes a lot of sense to make a few bodies with this feature.

 

[Tom_N]

It's not as if the manufacturer is going to offer to put a larger sensor into the DSLR or MILC that you have now.

Basically it is. Size of the sensor, shutter and displays + whatever changes to the electronics and proportions of the camera are needed but the lenses would be the same, only you would get more resolution and AoV out of them.

Even if you own a MILC, the manufacturer isn't going to offer to put an oversized sensor into the camera that you already own, for a low, low upgrade price.

You will need to buy a new body – a.k.a. forklift upgrade.

 

[Michael Fryd]

True that, but in another article explaining the manufacturing process seems to imply that is exceptionally difficult, so not realistic. The leap from rectangle to square seems more technologically possible.

I shall pray to the patron saint of photography......LOL

If you want the most flexibility, then produce a square sensor that completely covers the image circle, and take whatever cropping you want.

However, be prepared that the resulting image quality may not be as high as if the camera system was designed for your crop.

A typical full frame DSLR lens doesn't really project a round image circle. There are baffles inside the lens barrel and camera that restrict some of the light that would fall outside the 2:3 aspect ratio sensor. By reducing this stray light, you reduce veiling flare and increase the contrast of the final image.

In terms of image quality, cost, convenience, etc., the best sensor shape is the one that matches your final image.

 

[Iliah Borg]

There is no reason that square sensors can't be made,

And of course they are made. IMX533 is one example.

but the square format has never been as popular as the rectangular format. Back in the film era we had a variety of very fine twin lens reflex, Hasselblad and other cameras that shot square, but often as not they ended up getting cropped to a rectangle anyway.

 

[rurikw]

It's not as if the manufacturer is going to offer to put a larger sensor into the DSLR or MILC that you have now.

Basically it is. Size of the sensor, shutter and displays + whatever changes to the electronics and proportions of the camera are needed but the lenses would be the same, only you would get more resolution and AoV out of them.

Even if you own a MILC, the manufacturer isn't going to offer to put an oversized sensor into the camera that you already own, for a low, low upgrade price.

Oh, not sure how I missed that you meant it literally: sending an old camera for upgrade. I was thinking of new models within the old system.

You will need to buy a new body – a.k.a. forklift upgrade.

Not familiar with the concept of "forklift upgrades" but yes of course, we are talking about new camera bodies for old systems.

 

[Currantos]

I should have stated immediately that it should be oversized such as 18x18 for M43, 24x24 for APS and 36x36 for FF. Doesn't make sense otherwise. The corners of the sensor are wasted whereas you gain now wasted parts of the image circle. A clear win to me. I'm aware that FF is pushing the border of affordable sensor size so 36x36 might not be viable yet except as a medium format alternative with similar pricing but MF is coming down too. Another qualification is that it's not for DSLR as there is no space available for the bigger mirror. Nevertheless there's always been mirrorless where it's a logical solution. In the beginning of digital, sensors were probably much more expensive and made up a larger part of the price of the camera so it made sense to stick to film logic and statistics of format preferences thus saving crucial mm^2s of sensor area but that hasn't been the case for a long time now.

For about 50 years a very (? the most) popular style of camera was an SLR. This used the frame size of 36 x 24mm that became popular 100 years ago and still is.

To make a camera with film or sensor 36mm high would mean increasing the height of every one of those cameras. For (D)SLRs it would also mean that the body would have to be 12mm deeper to accommodate the larger mirror; and that would mean that existing lenses would be needed with a longer flange distance. In other words, not an economic proposition.

Moreover, to utilise the full area of the 36 x 36mm frame the image circle is 51mm* compared to 43mm so either:

(a) a whole new series of bigger, heavier (and more expensive) lenses would be needed or

(b) much of the area of the sensor would be wasted. In fact, the square to fit inside 43mm is just under 26mm so all you get is a fractional increase over the 24mm currently available.

* Note that many medium format digital cameras require 55mm image circle so the new camera would cost close to the cost of medium format digital, camera.

Please correct me if I am wrong.

To fit inside a 43mm circle the sides of the square would be 30.4mm, not 26mm. I do not understand where 26 comes from.

As for the area covered with pixels

30x30 =900 sq mm.

36x24 = 864 sq mm.

According to geometry the square "wins" against any rectangle so you always end up with more pixels total. I understand other things might make things more or less useful but from a point of capturing most light on most pixels the square wins according to Pythagoras.

This is not to win any debate or change anyone's mind, it's just that some people think different and would be interested in that idea, like myself. Not to prove me right or anyone else wrong or anything. I just really like the idea of a square for my shooting needs.

Cheers everyone, and don't be a square. They are not cool.

 

[Currantos]

r

As I said going beyond 36x24 might be expensive at the moment so the concept is probable more viable for APS. You can easily scale down the above chart, it works the same way. There is a small error in the diameter (44 vs 43.27) but that dos not affect the overall picture.

This is a fantastic demonstration.

I actually think, and maybe I am 'original' that 36mm square with corners cut off by hard vignette from the lens circle would be absolutely best. It would be so easy to just click in LightRoom on different possible crops and get rid of it. But the number of options and the total light captured and information would be the greatest. So you would have a larger number of choices of different cropping options because you would be again, choosing from a larger starting point.

Fantastic idea, too bad most people would be freaked out by it. Perhaps manufacturers could have an internal setting as to what crop you want when you shoot, but the entire square sans corners is captured and available to in case it is needed. So many times I cut off a hand or a foot on a person, especially athletes and dancers that move around like crazy. To have those extra 2-3mm of sensor real estate would be so awesome. Otherwise I have to keep on shooting wider and wider with lower and lower resolution.

 

[Christof21]

r

As I said going beyond 36x24 might be expensive at the moment so the concept is probable more viable for APS. You can easily scale down the above chart, it works the same way. There is a small error in the diameter (44 vs 43.27) but that dos not affect the overall picture.

This is a fantastic demonstration.

I actually think, and maybe I am 'original' that 36mm square with corners cut off by hard vignette from the lens circle would be absolutely best. It would be so easy to just click in LightRoom on different possible crops and get rid of it. But the number of options and the total light captured and information would be the greatest. So you would have a larger number of choices of different cropping options because you would be again, choosing from a larger starting point.

Fantastic idea, too bad most people would be freaked out by it. Perhaps manufacturers could have an internal setting as to what crop you want when you shoot,

Exaclty, this is really missing !!

but the entire square sans corners is captured and available to in case it is needed. So many times I cut off a hand or a foot on a person, especially athletes and dancers that move around like crazy. To have those extra 2-3mm of sensor real estate would be so awesome.

Yep, no need to have a 36x36 sensor, 36x30 would be awesome

 

[Michael Fryd]

...

Please correct me if I am wrong.

To fit inside a 43mm circle the sides of the square would be 30.4mm, not 26mm. I do not understand where 26 comes from.

As for the area covered with pixels

30x30 =900 sq mm.

36x24 = 864 sq mm.

According to geometry the square "wins" against any rectangle so you always end up with more pixels total. I understand other things might make things more or less useful but from a point of capturing most light on most pixels the square wins according to Pythagoras.

This is not to win any debate or change anyone's mind, it's just that some people think different and would be interested in that idea, like myself. Not to prove me right or anyone else wrong or anything. I just really like the idea of a square for my shooting needs.

Cheers everyone, and don't be a square. They are not cool.

Short answer:

From a scientific standpoint, if we ignore the esthetic intentions of the photographer, and we want to limit ourselves to a rectangular sensor, a square sensor can capture the most light.

From an engineering standpoint, if we want to maximize quality in the final image, we want a sensor aspect ratio that matches the aspect ratio of the final image.

Long answer:

It depends on your goal. If you goal is to find the rectangular with the largest area that fits within a specified circle, the square wins.

However, if your goal is to capture the most light for your image, then parts of the sensor that get cropped out don't count.

If you are willing to produce only square images, then a square sensor will provide the highest quality from a given image circle. On the other hand, if you are producing rectangular images, you get the best quality out of that same image circle by having the aspect ratio of the sensor match the aspect ratio of your final image. That way you don't lose any sensor area to cropping.

If you want an aspect ratio for your sensor that is best in the generic sense, you are best off with an aspect ratio that closest to the aspect ratio of generic images.

 

[Michael Fryd]

r

As I said going beyond 36x24 might be expensive at the moment so the concept is probable more viable for APS. You can easily scale down the above chart, it works the same way. There is a small error in the diameter (44 vs 43.27) but that dos not affect the overall picture.

This is a fantastic demonstration.

I actually think, and maybe I am 'original' that 36mm square with corners cut off by hard vignette from the lens circle would be absolutely best. It would be so easy to just click in LightRoom on different possible crops and get rid of it. But the number of options and the total light captured and information would be the greatest. So you would have a larger number of choices of different cropping options because you would be again, choosing from a larger starting point.

Fantastic idea, too bad most people would be freaked out by it. Perhaps manufacturers could have an internal setting as to what crop you want when you shoot, but the entire square sans corners is captured and available to in case it is needed. So many times I cut off a hand or a foot on a person, especially athletes and dancers that move around like crazy. To have those extra 2-3mm of sensor real estate would be so awesome. Otherwise I have to keep on shooting wider and wider with lower and lower resolution.

But if you want a 2:3 aspect ratio, a 30mm square sensor is not ideal. You end up with 20mm by 30mm of usable sensor compared to 24mm by 36mm in a traditional full frame camera.



If your 24x36mm full frame had 24 megapixels, than your 20x30mm sensor has only about 16.7 megapixels.

From a practical standpoint, if the sensor is contained in the image circle, the best aspect ratio is the one that matches your final print. That's the only solution where the corners of the image map to the edge of the image circle.

 

[rurikw]

r

As I said going beyond 36x24 might be expensive at the moment so the concept is probable more viable for APS. You can easily scale down the above chart, it works the same way. There is a small error in the diameter (44 vs 43.27) but that dos not affect the overall picture.

This is a fantastic demonstration.

I actually think, and maybe I am 'original' that 36mm square with corners cut off by hard vignette from the lens circle would be absolutely best. It would be so easy to just click in LightRoom on different possible crops and get rid of it. But the number of options and the total light captured and information would be the greatest. So you would have a larger number of choices of different cropping options because you would be again, choosing from a larger starting point.

Fantastic idea, too bad most people would be freaked out by it. Perhaps manufacturers could have an internal setting as to what crop you want when you shoot, but the entire square sans corners is captured and available to in case it is needed. So many times I cut off a hand or a foot on a person, especially athletes and dancers that move around like crazy. To have those extra 2-3mm of sensor real estate would be so awesome. Otherwise I have to keep on shooting wider and wider with lower and lower resolution.

Yes I think getting a larger angle of view and more pixels every time you make a photo with a more square ratio than the old 3:2 or 4:3 aspect would be a great asset. And it would actually be economically sensible as many of us spend more money on lenses than bodies.

 

[Tom_N]

You will need to buy a new body – a.k.a. forklift upgrade.

Not familiar with the concept of "forklift upgrades" but yes of course, we are talking about new camera bodies for old systems.

It's more of a computer/IT term than a photography one.

It refers to an "upgrade" that requires a massive replacement of equipment, so massive that there there would likely be forklifts going into your machine rooms … to carry away heavy pallet loads of old hardware and to bring in heavy pallet loads of new hardware.

Forklift upgrades are said to be good for the vendor (KA-CHING!), but not always as good for the customer.

 

[fotowbert]

There is no reason that square sensors can't be made, but the square format has never been as popular as the rectangular format. Back in the film era we had a variety of very fine twin lens reflex, Hasselblad and other cameras that shot square, but often as not they ended up getting cropped to a rectangle anyway.

The square format has the advantage that you didn't have to commit to landscape or portrait orientation while shooting. That decision can be made latter during editing/printing but at the expense of some image area.

As you say I almost always cropped my 6X6 Mamiya C330 images to rectangular proportions during printing years ago.

• John