[emalvick]

thats if all light were equal, but its not its a spectrum (a condition that is not limited to a specific set of values but can vary infinitely within a continuum)

for every element you add to correct one aberration, that element introduces another aberration, this is why modern prime lenses can have upto 11+ elements and cost more than most camera bodies

I'd still argue it's easier to hit a flat plane. The fact that its a spectrum is irrelevant other than it makes bending light more difficult, and hitting a curve would be even more difficult. Plus, I'd argue that modern lenses have it down pretty well. Aberrations can and do occur but usually for historic lenses and they are mostly a result of the sensors and digitization not necessarily the lenses. Thus, older lenses that show aberrations on our dSLRs had no visible aberrations on film.

Afterall, while light is a spectrum, the visible spectrum that lenses have to capture is not nearly so wide, is definitely finite, and not that difficult to control by modern optical engineers. Most of the infinite light spectrum is not visible light and is irrelevant to the optics industry.... yes despite your need to define what a spectrum is on a photography forum, it might surprise you that most people already know what it is. And, despite the complexity a spectrum makes, I don't think it changes my argument one bit.

I mean, if you think about the implications of trying to design for a curved sensor, imagine the complexity of a zoom lens, which can vary in focal length as the light spectrum would.

The reality is, a plane is mathematically easier to work with even if it isn't the ideal. If anything, the complexity of what could be achieved may be more achievable through post-processing.

[Ratcheteer]

I'd still argue it's easier to hit a flat plane. The fact that its a spectrum is irrelevant other than it makes bending light more difficult, and hitting a curve would be even more difficult.

first off bending light always leads to an aberration of some sort we are dealing with multiple wavelengths not all will end at the same point hence why i brought up the spectrum,

second, bending light to hit the edges of a flat plane means that it hits at an angle, not straight on (this is what causes the most visible aberrations - vignetting and color shifts, coma, chromatic aberration.)

third, most lens made today have a curved field curvature (not flat as some may believe) including the very popular Distagon and the batis loxia touit

"distortion free vignette free sharp edge to edge 8k video" with a cheap flat sensor, cheap (but complex) lens, and software.

i work with digital cine cameras capible of shooting 8k and lately I've been using Zeiss Milvus lenses and they still have some vignetting, you talk of filters, i call that grading which i need a render farm if I'm to spit out (33megapixel) 43,200 rawDNG frames (30mins@24fps) in under 30mins, somehow i dont see this sort of power coming to a cellphone near you anytime soon, where would they put the 16 lens elements and this is without auto focus

the samsung s7 only has 12 actual megapixels but the quality is closer to 4 perceptual megapixels

[emalvick]

first off bending light always leads to an aberration of some sort we are dealing with multiple wavelengths not all will end at the same point hence why i brought up the spectrum,

second, bending light to hit the edges of a flat plane means that it hits at an angle, not straight on (this is what causes the most visible aberrations - vignetting and color shifts, coma, chromatic aberration.)

third, most lens made today have a curved field curvature (not flat as some may believe) including the very popular Distagon and the batis loxia touit (the proof is in the bokeh)

I never questioned any of that is true; the physics aren't in question. But each focal length will have a different curvature. How do you remedy that? How do you deal with the fact that interchangeable lenses have been engineered for years often with minor changes, thus the economy of what we have

Mathematically, it's a lot easier (not necessarily better optically) to design for aberations, distortion, etc when you are trying to project onto something flat. If you can only curve a sensor for one ideal, every other lens has to design for the curve of its field of curvature and project that onto another curve, which will generally not be the same.

Aberrations will still occur and so forth. And, I'd argue that while aberrations occur, they are not nearly that bad. The only lenses where I see aberrations as problematic are older lenses that come from a period when we didn't have pixels.

Again, while a curved sensors sounds ideal, for an interchangeable lens camera, I only see it as a huge cost and an design challenge that either requires the sensor to change curves and all new lenses, which will cost a lot pending the exactly how the sensors are designed. Maybe it'll be practical in the future. I'm sure it is practical now for a fixed focal length camera.

[Ratcheteer]

But each focal length will have a different curvature. How do you remedy that?

forget field curvature for a minute, the reality is lens designers consider this a good trade off for lens sharpness since its not all that visible if your not looking for it in the first place and in fact zoom lenses often trade one or more aberrations for higher image sharpness.

light also disperses, even a focused beam will lose intensity as the beam will get wider over distance, you might not think this is a big deal but that is only one of many factors that is causing peripheral fall off and drop in contrast at the edges of the frame

the one i believe that causes the most problems is angle of light and is also why i think your argument for curved sensors only benefiting fixed focal length cameras is a bit closed minded.

take a torchlight shine it on a flat piece of paper from 4 inches away at a dead on angle you just a nice round patch illuminating the paper evenly, the moment you angle the light the patch is no longer round now nor evenly illuminated, repeat the same test from different distances 2 inches away and 6 inches away (this is like zooming in and out) the results will be the same (this test is actually how you demonstrate the effect of a aberration known as coma)

now repeat the test with a empty fruit bowl and right away you will see the benifit the moment you change the angle, of course one distance will work better than the other but they all are receiving the benefit of illuminating a curved surface (the size/shape of the patch is better controlled and there is less fall off from one edge of the patch to the other vs the paper test.

now back to field curvature, yes zoom lens has varying curvature but its far away from being flat, but the fruit bowl results should show you that a curve surface can in fact benefit zooms, it might be only a little but any thing that helps flatten the curvature has a list of other beneficial effects on image quality (coma resistance, lower peripheral fall off (vignetting), reduce edge astigmatism) those very benefits would allow for future lens to require far less elements, most existing zoom lenses do not correct for the benefits a curved sensor could offer anyway, the only incompatible lenses would be high quality primes :P