Patent: Electronic Curved Sensor

[ExodistPhotography]

How many Bendings/Stretchings will such a sensor survive? How thin must a bendable sensor be to allow efficient curvature? What are the dimensions we are talking about: how many µm (?) will the Corners be moved foreward in relation to the Center of the sensor? Will a bendable sensor be permanently curved or only during exposure? Will curvature be individually adusted, just like AFMA? Can curvature of a sensor be used to adjust slight decentering?

It will likely be so minimal that you would not be able to tell it just from looking at the sensor. This is something similar to what you can do with reflecting "Newtonian" telescopes were the main mirror is bent to fine tune the telescope's optical performance.

 

[melgross]

A curved sensor doesn't really solve any problems and makes things worse..... but a bendable curved sensor, now that's different! leave it flat for traditional lenses, or bend it to the focal length curvature of a new lens design.

This would allow the lens designer to NOT have to try to bend the light to approximate a flat plane, and that means fewer elements, which should translate into less lost light, sharper images, and lower cost. This is a really big thing!

From Canon's viewpoint, it would, of course, only be bent to match Canon lenses, so 3rd party lenses would lose out.

Still a big win, since it allows both 'old' lenses (flat field, at least designed that way) and 'new' lenses (not corrected for field curvature) to be used on new bodies.

I expect 3rd parties would approximate which Canon lenses theirs most closely matched, and spoof that LensID, much like they do now for accessing the AF algorithms.

I don't see it as being a problem in that way. The first thing to note is that if only Canon is doing this, it's unlikely that a third party lens manufacturer would bother. Right now, optical designs are the same, and it's the minor difference of the lens mount and electronics that they match. Since they don't do that for all mounts, because of sales, can you imagine them designing lenses just for Canons' platform? That would be a difficult decision.

The second thing is that if they did decide to do it, they wouldn't need to exactly duplicate it. If there's a big enough difference, optically, between the curved sensor and the flat state, then just approaching it would result in a big improvement.

Thirdly, since it appears that the curved state is functioned by an electric signal, it's possible that it can vary its curved state depending on the signal. Perhaps wide lenses could see more curvature, and tele's, less. It's even possible that the curve could vary according to the focal length in a zoom. Now, that would be something!

Lastly, what's to say that Canon would even allow the curved state to work with third party lenses?

 

[melgross]

Could a bendable sensor be able to reduce vignetting on current lens - especially for wide angle?
Would a body with a curveable sensor be limited by its resolvable resolution ie lower pixel count as getting a bendable sensor relax into a perfectly flat default position may be challenging.

No. A bent sensor will not work with any EF lenses. The design will be for fixed lens cameras both fixed focal length and zooms (variable bend), things like the security cameras and industrial cameras that Canon have said they are keen to get in to.

I wouldn't be so quick to make a dramatic statement like that. There are very few lenses that are truly designed for a real flat field. Process lenses, which I had in my halftone camera, was one of the few. But no matter what they may say, no "normal" lens is a true flat field lens. Digital lenses are closer to that than film lenses were, but not completely. Otherwise, many aberrations would be a memory. In my one year of optics, many years ago, we were shown that even a tiny amount of film (sensor today, of course) curvature would make a major improvement in picture quality. But it was too difficult to make cameras that curved the film plane, that weren't much more complex, and didn't put too much tension on the film in winding.

So it's possible that Canon would start with speciality equipment, and with experience, and lower costs, would bring it to the the regular picture taking market. But they might also need new lenses to take best advantage of it, and that's more time and money.

 

[Loswr]

I don't see it as being a problem in that way. The first thing to note is that if only Canon is doing this, it's unlikely that a third party lens manufacturer would bother. Right now, optical designs are the same, and it's the minor difference of the lens mount and electronics that they match. Since they don't do that for all mounts, because of sales, can you imagine them designing lenses just for Canons' platform? That would be a difficult decision.

The second thing is that if they did decide to do it, they wouldn't need to exactly duplicate it. If there's a big enough difference, optically, between the curved sensor and the flat state, then just approaching it would result in a big improvement.

My point was not that 3rd parties would design new lenses for a curved sensor, but rather that they would not need to change their flat-field designs to work with such a curved sensor.

Thirdly, since it appears that the curved state is functioned by an electric signal, it's possible that it can vary its curved state depending on the signal. Perhaps wide lenses could see more curvature, and tele's, less. It's even possible that the curve could vary according to the focal length in a zoom. Now, that would be something!

I believe that's the whole point of this patent - a dynamically controlled amount of curvature. The necessary degree of curvature to avoid the need for in-lens optical correction is going to vary based on focal length, so a fixed sensor curvature would only be optimally effective at a single focal length. Without the ability to dynamically adjust the curvature, the application would be limited to a fixed prime lens camera, not a fized zoom lens or an ILC.

Lastly, what's to say that Canon would even allow the curved state to work with third party lenses?

Good point, but by the same token why should they allow a 3rd party lens to work with their AF system? But...they do.

 

[melgross]

I don't see it as being a problem in that way. The first thing to note is that if only Canon is doing this, it's unlikely that a third party lens manufacturer would bother. Right now, optical designs are the same, and it's the minor difference of the lens mount and electronics that they match. Since they don't do that for all mounts, because of sales, can you imagine them designing lenses just for Canons' platform? That would be a difficult decision.

The second thing is that if they did decide to do it, they wouldn't need to exactly duplicate it. If there's a big enough difference, optically, between the curved sensor and the flat state, then just approaching it would result in a big improvement.

My point was not that 3rd parties would design new lenses for a curved sensor, but rather that they would not need to change their flat-field designs to work with such a curved sensor.

Thirdly, since it appears that the curved state is functioned by an electric signal, it's possible that it can vary its curved state depending on the signal. Perhaps wide lenses could see more curvature, and tele's, less. It's even possible that the curve could vary according to the focal length in a zoom. Now, that would be something!

I believe that's the whole point of this patent - a dynamically controlled amount of curvature. The necessary degree of curvature to avoid the need for in-lens optical correction is going to vary based on focal length, so a fixed sensor curvature would only be optimally effective at a single focal length. Without the ability to dynamically adjust the curvature, the application would be limited to a fixed prime lens camera, not a fized zoom lens or an ILC.

Lastly, what's to say that Canon would even allow the curved state to work with third party lenses?

Good point, but by the same token why should they allow a 3rd party lens to work with their AF system? But...they do.

It's rather interesting. I would love to read the actual patent myself, but I haven't seen it yet. Again though, if the difference between the states is minor enough, it may not matter. A reason for that is that lenses are now designed, intemtially, or not, with a bit of curvature. The curvature will follow the curve of the sensor, whether or not it matches, or not. So you could very well be correct about lenses benefiting even if they aren't designed for it.

But if they are, I can see a major improvement, not in falloff, as that's still more of a front element size thing than anything else, but in most other aberrations.

The problem is whether these new lenses would perform worse on flat sensors. If so, then that would be a marketing problem, as most all cameras will still have flat sensors. What would be interesting are lens elements that themselves change shape with an electric charge. Phillips was working on such a thing around the year 2,000, if I remember correctly. That could also fix these problems, and work with any sensor.