Have we reached ‘peak’ image quality?

[Mr Badger]

Right. That's all we wanted to know. The rest is irrelevant.

So I'd need to actually own a Ford Fiesta 1.1L to know that a Ferrari 633 was quicker then? Does me not owning a D600/610 somehow magically improve it's available ISO range and signal to noise ratio, and improve its dynamic range above 800 ISO? No, it does not!

 

[AZ6]

So I'd need to actually own a Ford Fiesta 1.1L to know that a Ferrari 633 was quicker then? Does me not owning a D600/610 somehow magically improve it's available ISO range and signal to noise ratio, and improve its dynamic range above 800 ISO? No, it does not!

What on EARTH are you wibbling on about now? Ford Fiestas? Forgive me; I've already forgotten what you were talking about anyway. It was that interesting. Why aren't you taking loads of amazing pictures with your wonderfully superior camera?

 

[Mr Badger]

More to the point, why aren't you discussing whether or not we've reached peak image quality?

________________________________________________________________________​

I suspect as time goes on there will continue to be incremental improvements in ISO performance, dynamic range, resolution, etc. Perhaps also improvements to file types to reduce compression loss, to make file sizes more manageable as the resulting pixel count continues to grow?

It's probably hard to imagine something much better than the Nikon 850, top of the range Sony mirrorless, etc. but it will be out there eventually. Who knows, in 20 or 30 years time zooming in to 400% may look like 100% is now in terms of image resolution and detail?

Also I think further improvements to features such as focus point selection and distribution across the viewfinder/screen will occur. Perhaps even a return of eye controlled focus point selection too, so you have the option to just look where you want to focus, instead of joysticking or button scrolling?

I think camera manufacturers are going to have to keep improving and innovating if they are to keep the market alive against competition from top-end smart phones. That's already pretty much killed the 'point and shoot compact' end of the market, and multi-lens smart phones are now eating away at the entry level DSLR and mirrorless end of the market. If manufacturers don't keep working to raise the bar with better image quality for enthusiast and pro grade cameras, then I don't think they'll survive.

I take the point about 'how good does it need to get', after all, digital cameras from 10 or 20 years ago can still produce some very nice looking photos, until the light gets tricky and/or you zoom in to the image, then the difference becomes apparent.

 

[Joel_Partch]

Perusing some very sharp images from my D810/D850 I cant imagine that anyone would want/need better iq nor could ‘better’ be easily perceived/resolved by the human eye.

Whilst af fine tuning my D850 today with a 70-200 I had to go to 300% to see the difference it was making.

Whilst a computer maybe able to split hairs amongst the best lenses the human eye simply isn’t that good.

So have we reached ‘peak’ image quality?

You could argue the D800E was the moment IQ peaked: it was a big step up from anything else available previously. The D850 is a very minor incremental improvement, and nothing since then has been such a large step up. Ironically, the biggest jump over the D800E was probably the ancient Canon 5DSr, which might still be the best thing out there in terms of single-frame image capture.

The trend toward lower-resolution sensors - and pixel shifting - that give the look of full frame without the pixels is tacit admission that we've bounced off a brick wall.

It's all about the smarts now: AF has come on in leaps and bounds since then and we're on the brink of seriously useable computational photography. In-camera lens correction is only the first step. The move toward mirrorless was inevitable, given that it opens the door to improved lens design, which is another area that has taken great strides. But if your photography doesn't involve speed, there's barely any need to upgrade from a Nikon D8XXX or Canon 5Ds - just attach newer, better lenses.

 

[Brazo]

You could argue the D800E was the moment IQ peaked: it was a big step up from anything else available previously. The D850 is a very minor incremental improvement, and nothing since then has been such a large step up. Ironically, the biggest jump over the D800E was probably the ancient Canon 5DSr, which might still be the best thing out there in terms of single-frame image capture.

The trend toward lower-resolution sensors - and pixel shifting - that give the look of full frame without the pixels is tacit admission that we've bounced off a brick wall.

It's all about the smarts now: AF has come on in leaps and bounds since then and we're on the brink of seriously useable computational photography. In-camera lens correction is only the first step. The move toward mirrorless was inevitable, given that it opens the door to improved lens design, which is another area that has taken great strides. But if your photography doesn't involve speed, there's barely any need to upgrade from a Nikon D8XXX or Canon 5Ds - just attach newer, better lenses.

I’m a big advocate of computational photography but I think it will only give us what we already have (iq wise) albeit in a smaller, easier to carry (mobile phone!) package, or indeed some flashy computational hybrid camera with multiple lenses and perhaps multiple sensors but still compact in size.

 

[StephenM]

Two ways of looking at the original question for me, and neither involve low light capability (since the question did not specify the genre of photograph) or sensor noise (as that also doesn't affect me).

Could my technical image quality be improved beyond what is currently possible with existing equipment? Yes, because better lenses could presumably be computed and made for 20x16 cameras. So, no, we aren't at the peak of what is possible. But I doubt very much whether that will happen. So I'd say imrovements are possible, so we aren't at the peak. But improvements are unlikely to be made, so we are at the peak.

And if this seems off topic, digital wasn't specified

 

[AZ6]

And if this seems off topic, digital wasn't specified

I was wondering recently, whilst perusing some old slides and negs, how much 'better' my current digital cams are at producing pics than my old film gear. Because with film, you first have that initial chemical reaction, as the film is exposed, then another as it's developed, then you print using an enlarger, so another lens is put into the mix. At each stage there is potential for generational degradation, surely. Whereas with digital, you can see the pic as the camera's sensors saw it; I spose you have the 'generational' step with the computer rendering it on your screen. But my digital images with the same lenses, seem just that bit sharper. It would be interesting to do some proper side by side tests mind. But that either involves printing out a digital image to match a film one, or scanning in a neg/transparency. So you'd have that generational step either way.

Hmm. I have the equipment to actually do this...

 

[Neilmack]

The law of diminishing returns is setting in with a vengeance. Once we looked forward to digital cameras that could produce an adequate 5"x7" print. Now cameras with 40megapixel plus resolutions are everyday, complete with huge dynamic ranges, and credible performances at up to 2000 ISO. The latest iterations of several successful cameras have actually reduced their pixel count. If I replace my D800 with a D850, it won't be image quality per se that forces the issue - it'll be handling considerations like a hinging rear screen.

 

[AndrewFlannigan]

So I'd say imrovements are possible, so we aren't at the peak. But improvements are unlikely to be made, so we are at the peak.

Several years ago there were reports about work on an offshoot of holography that would capture the wavefront of a selected area of light using interference patterns. To get an idea of what was proposed you need to think about the Williams–Kilburn tube used as random access storage in some computers in the late 1940s (see here).

In the imaging proposal a scanning plate would be the sensor and the storage would be loaded by passing an access beam across the sensor and onto a receptor which would store the image bytes into RAM. In theory, the definition would be limited entirely by the fineness and tracking accuracy of the scan beam. Some articles claimed this would result in detail down to an almost microscopic scale (think resolutions in the Gigapixel range).

I think that the last time I saw anything about it was around twenty years ago so I don't think silicon wafers are in danger of being superseded any time soon,