[Belcik]

hi,

Out of being tired (doing godzillion things at the same time). Two questions came into my mind.

1. Why there is no round image sensors and round viewfinders? It seems that we are to much used to past...
2. Why image sensors are flat? I mean, lenses deliver picture that is compressed at edges. Instead of struggling with corrections, wouldn't it be more reliable to have slightly bent inside sensor?

I swear, i did not drink and smoke anything. These two things just popped into my mind and I could not find resanoble answers for them...

[kavolis]

round sensor? Here you go: Nokia 808 PureView - Wikipedia, the free encyclopedia

[TER-OR]

I would think different lenses would deliver different-shaped curved plane - or sphere-section, I suppose - images so one sensor curvature may not work for different lenses - or maybe even not within zoom ranges on one lens. So each lens would still need some correction - and that would need to be standardized.
Of course a dedicated camera/lens format could do this. I know some groups have been playing with making true spherical lenses work - though they have limitations.

It of course all does stem from a planar film as recording media.

Nuthin' wrong with thinking, though. Maybe some web searching for spherical lenses or curved sensors would yield some information.

[SOldBear]

1. Why there is no round image sensors and round viewfinders? It seems that we are to much used to past...
2. Why image sensors are flat? I mean, lenses deliver picture that is compressed at edges. Instead of struggling with corrections, wouldn't it be more reliable to have slightly bent inside sensor?

Disclaimer: there is nothing scientific about my post, just feelings.

- Ultimately, the image is viewed on a flat and rectangular media (books, monitor screens, billboards, brochures, ....).

- Let's say image sensors can be made round and non-flat. Then at some point in the process, the image has to be flattened and cropped to rectangular. We may as well start with flat and rectangular.

- I don't think the image sensors is manufactured one by one, instead, they are made many on a large "sheet," and then cut to individual sensors, similar to the way many electronic chips are made from a large wafer. This way of manufacturing can't make round and non-flat sensors.

---

Last edited by SOldBear; 09-19-2012 at 09:04 PM.

[noblepa]

I, too, have wondered if a spherical-section sensor wouldn't reduce distortion. But, as TER-OR suggested, different focal lengths might require a different spherical radius.

It would also be pretty expensive to manufacture a spherical sensor, I would think. Current manufacturing techniques involve imprinting as many sensors as will fit on a round disk, and then cutting them out, like cookies. That's one reason often given why FF sensors are more expensive than aps-c sensors. Fewer FF sensors will fit on the disk.

Spherical sensors would have to be manufactured individually, which would blow the cost way up. At the very least, an entirely new manufacturing technique would be required.

[Lowell Goudge]

hi,

Out of being tired (doing godzillion things at the same time). Two questions came into my mind.

1. Why there is no round image sensors and round viewfinders? It seems that we are to much used to past...
2. Why image sensors are flat? I mean, lenses deliver picture that is compressed at edges. Instead of struggling with corrections, wouldn't it be more reliable to have slightly bent inside sensor?

I swear, i did not drink and smoke anything. These two things just popped into my mind and I could not find resanoble answers for them...

If you draw a series of (for sake of argument) 28.8mm circles on a 200mm diameter circle, look at all the scrap. Now do the same with 16 x 24mm rectangles. The packing factor is better with the rectangles, better packing factor, better utilization of the cilicon crystal, whihc is expensive. If the biggest crystal they could grow is 29mm then you might have an argument

as for flat vs curved, the sensors are sliced out of a crystal, and are NOT flexible, the structure is flat by nature of the atomic links. curved surfaces dont work easily because they disrupt the crystal structure.

[lister6520]

Curved sensors do indeed exist, but they are extremely rare, being designed and built for one of a kind instruments typically special purpose astronomical telescopes and some unusual military imaging systems.

They are extremely expensive to manufacture, expensive as in selling the house won't pay for it. The justification for their existence is that they allow the design very wide aperture lenses (under F1) which is something practically impossible to achieve in combination with a flat image field.

[jimr-pdx]

hi,

1. Why there is no round image sensors and round viewfinders? It seems that we are to much used to past...

Personally I'd love a cam with a 25Mpx square sensor, from which I can pick the aspect ratio. That would include portrait orientation while holding the camera in landscape position - I can't explain why that's a really stupid idea, but someone will likely educate me. And like you I've not been drinking either, though I've a touch of ibuprofen in my system.. and I may be a bit feverish?

[ironlionzion]

look around you - see any aberrations in the image in your brain? Nope. That's because the curved "sensor" in your eye naturally corrects for them. The problem, as stated previously, is constructing curved sensors cheaply.

The fact that we are eventually viewing the image on a 2D surface has nothing to do with the sensor. One of the most significant problems lens designers face is the fact that the distance of each pixel in the sensor to the center of the lens increases as you move from the center of the sensor. A curved sensor mitigates this problem.

[RKKS08]

The Minox ("spy") cameras were designed for repro work, and specially macro repro work. For this purpose they would have needed a flatfield lens, but at that time there was none available.

So, up to the Minox B, they used an ordinary (curved field) lens. They bent the film instead, which could easily be done with high precision (8x11mm format!). Only with the Minox BL they introduced a flatfield lens and skipped that special feature.

In the early eighties, I did own both the Minox B and the BL, and it was described in the manuals (and you could see the curvation when loading/unloading film).

As others stated before, mass production of curved sensors would be extremely expensive.

[Ignacio.at.BA]

Hi Belcik & al.,
I'm pretty sure you're not talking about this... but some people are working in a very interesting light-field camera ( Light-field camera - Wikipedia, the free encyclopedia ), and at this moment somebody could buy a Litro one.

In the other hand, it seems to be some research on light fields and computational photography, in a campus "similar" to the one you noted, as can be read in: Light fields and computational photography (look for Mark Levoy's home page).

I know those publications are not about round sensors, buy I imagine strong changes on the way we'll be used to take pictures in a future.
Regards

Ignacio

[Stone G.]

Curved sensors do indeed exist, but they are extremely rare, being designed and built for one of a kind instruments typically special purpose astronomical telescopes and some unusual military imaging systems.

They are extremely expensive to manufacture, expensive as in selling the house won't pay for it. The justification for their existence is that they allow the design very wide aperture lenses (under F1) which is something practically impossible to achieve in combination with a flat image field.

Not completely correct. Many of us from the analogue ages may remember the (film) Schmidt camera used for astrophotography:

Schmidt camera - Wikipedia, the free encyclopedia

Schmidt cameras use(d) simple, sperical mirrors that are in fact easier and cheaper to make than higly corrected parabolic or hyperbolic mirrors. And for the amateur astrophotographer, you could actually buy small schmidt cameras that came with 35mm film holders which squeezed the film into a circular, curved shape that matched the curvature of the focal sperical plane of the mirror.

So there you go: Circular and curved sensors do exist in a certain sense. But yes, a curved CCD or CMOS would indeed be a costly item with current day's manufacturing technology.

[cbaytan]

hi,
1. Why there is no round image sensors and round viewfinders? It seems that we are to much used to past...

If you print or look them round that would be ok, otherwise you will be cropping to rectangle anyway.

hi,
2. Why image sensors are flat? I mean, lenses deliver picture that is compressed at edges. Instead of struggling with corrections, wouldn't it be more reliable to have slightly bent inside sensor?

This is the similar question I was asking myself years ago, I was thinking a curvature changing sensor instead, as we changed the focal length (yeah right! carry on babe ) , but this way lenses would be lot simpler.

So, what you are saying is possible *if* camera has a particular fixed single focal point lens, so sensor curvature would fit the lenses curvature all the time. But if such a sensor exists lens could be lot lot simpler, like 2 elements design or something.

[tim60]

hi,

Out of being tired (doing godzillion things at the same time). Two questions came into my mind.

1. Why there is no round image sensors and round viewfinders? It seems that we are to much used to past...
2. Why image sensors are flat? I mean, lenses deliver picture that is compressed at edges. Instead of struggling with corrections, wouldn't it be more reliable to have slightly bent inside sensor?

I swear, i did not drink and smoke anything. These two things just popped into my mind and I could not find resanoble answers for them...

These are the questions from which interesting innovations come.

But:
1. The only thing like it was the round fisheye, and the old style faded edge print. People seem to like the image capture to be technically good across a rectangular space.
2. In the old days film rolled across the exposure area, and flat was practical. The alternative would have been rounded along the film and flat across it - very difficult geometry which could not have been workable with spherical ground lens elements.
Not, the sensor is a silicon device, and silican wafers are manufactured flat becasue of all the process steps required and the need to manufacture at a price private people could afford to buy.

PP obviates the benefit of your first idea.

Maybe in the future when someone works out a cheap way and an advantage to exploit the idea ...

[Belcik]

Hi You all,

Asking this question I felt totally stupid, but it seems that not only me had such a thoughts. It came to me upon working on theories called "Functional Fixedness". I just thought for a second - why in digital cameras the sensor is made the way it is. And I came to conclusion, that initially no one thought about anything else. They simply copied the regular film.

But true, it would be very interesting, having the sensor bending a little bit depending on the focal length. That would be somewhat of revolution. But to be honest- what would be the real advantages of round bending (if round, then most probably collapsing) sensor? The obvious one is that you cover full Lens image, that you avoid some distortions. But it seems that it is all... no reason to reinvent that. True?