[WPRESTO]

It occurred to me:
One problem with maintaining IQ - particularly sharpness - to the edges and corners is field curvature of the lens, a problem that generally gets worse for wide angle lenses, the wider the worse the problem. This is an inherent problem of the property of optics and also the fact that objects being recorded at the edges of a WA image are commonly a greater distance away from the camera than objects near the center of the FOV. Would it possible to make a curved sensor, turned forward a minute amount at the edges, or even better, a concave sensor, with the corners curved forward slightly more than the midpoints along the edges of each adjoining side? Aside from providing some correction for field curvature of WA lenses, this would also align each pixel more directly with the path of the light it is recording, in a manner similar to the orientation of pixels at the center of the sensor. Would that slightly reduce vignetting? On the other hand, would such a curved sensor degrade the edges of images taken with long telephoto lenses, making such a concave sensor a more-loss-than-gain change?

[Ratcheteer]

I've brought up the topic of curved sensors as a comment on petapixel and the general consensus thought i was crazy.

but with megapixel count going up with the side effect of increasing diffraction limits, it doesn't surprise me others are starting to think of curved sensors too...

[WPRESTO]

Who ever imagined concavely-curved TV screens? So many years I watched a TV with a convexly-curved surface, then flat, not concave. Maybe it could be done with a sensor, but perhaps the manufacturing process precludes it. But would it actually be an improvement?

[gaweidert]

Since the image the lens projects is circular, and the sensor is rectangular curving the sensor would not be an easy thing. Some lenses are called flat field for a reason. A curved sensor may make it worst in that case. No matter what pixel peepers like to think, human perceptions has not changed and the majority of the population doesn't really care about the extreme edges of a picture. The eye is drawn to the center and it is actually how we ourselves see the world. I suppose it is technologically possible, but I really don't see a reason for on mass produced cameras. It would not surprise me is there are certain specific applications where curved sensors or a curved sensor array are already used. Much like some of the latest astronomical telescopes.

A curved TV screen is a whole other matter. Since the modern TV flat screen consists of pixels, the image is already as focused as it is going to gt everywhere on the screen. Curving it makes no difference. They curve a TV screen to more approximate how the eye sees the world. It may enhance the visual experience as the distance from the screen to the eye will be more constatn from center to edge. But I can imaging that the curve of the screen also forces a certain viewing distance for maximum effect.

[Ratcheteer]

Since the image the lens projects is circular, and the sensor is rectangular curving the sensor would not be an easy thing.

I think you have missed the point, its because the lens image is circular we get vignetting

but a better reason to use a curved sensor would be optics (who needs extreme rectilinear correction when the sensors curvature corrects the image for you)

[WPRESTO]

Yes, but, losing edge/corner IQ if there's a combination of curved sensor and flat-field lens would be one of the trade-offs. Most SFL lenses of around normal focal length, telephoto lenses, and macro lenses either attempt to be flat-field or are near flat-field because of inevitable optical properties (esp. with longer telephoto lenses).

[Ratcheteer]

Yes, but, losing edge/corner IQ if there's a combination of curved sensor and flat-field lens would be one of the trade-offs. Most SFL lenses of around normal focal length, telephoto lenses, and macro lenses either attempt to be flat-field or are near flat-field because of inevitable optical properties (esp. with longer telephoto lenses).

something that annoys me about camera manufactures the lens they make for their own bodys are only mildly corrected, the rest of the correction happens in body.

[dms]

Curved sensors would make the optics simpler--after all the reason for multi elements is largely to make the image rectangular--the image of a simple lens is a segment of a sphere's surface. The human eye does quite well this way. But in terms of practicality (i.e., impractical) perhaps this means a single FL lens--sensor (camera) combination.

[Ratcheteer]

But in terms of practicality (i.e., impractical)

probably not today but in the future I'm sure there will be medical uses for such a sensor

[WPRESTO]

Of course ground-based large astronomical telescopes now have "adaptive optics" that compensate continuously for atmospheric distortion. Image individual actuators and small patches of pixels that would move them in micro increments to achieve the best focus? Out of the question because the scale is way too small.

[Na Horuk]

The problem become things like different curvature for different lenses, register distance (backwards compatibility)

[biz-engineer]

Curved sensor is a good idea, but it would only be fit for wide angle lenses having the corresponding field curvature. The other way around: lenses field curvature allow sharper edges of indoor architecture images because the distances of edges is closer to the camera than the subject distance in the center. In other words , field curvature can be a controlled design feature for shooting architecture. The DA15 ltd yield very good edge sharpness for certain indoor photographs, and is a terrible lens for landscape.

[Barry Pearson]

something that annoys me about camera manufactures the lens they make for their own bodys are only mildly corrected, the rest of the correction happens in body.

For "happens in body" read "in post-processing".

I'm happy to have a lens with aberrations that can be corrected well in post-processing, (for example Lightroom), if that makes it easier to design the lens for other benefits, such as cost, weight, minimised aberrations that can't be corrected in post-processing, etc.

[RobA_Oz]

Sony was reported to have a full-frame (presumably 35mm) curved sensor camera ready to announce last December, but Christmas seems to be late there. Nikon was reported to have a curved-sensor compact in the works a few months ago, but we're still waiting on that, too.

Apple apparently has a patent on a curved sensor, too - Sony's is for a flat-formed sensor fixed to a dished ceramic backing, so I'm not sure what Apple's patent does differently.

It seems to be the next big thing, although that was also going to be Lytro, once, I recall. When Sony announced its patent in 2014, I recall some of our members using that to beat Ricoh around the head for not being innovative, because that was going to be the next big thing. Maybe it will be, but we're still waiting...

[Unregistered User 8]

Would a curved sensor have to " uncurve " as the lens was stopped down?