Lens resolution and large pixel counts

[Monobod]

Hi,

There is talk about the ability of medium priced lenses to register enough information to take full use of high pixel count sensors. I have noticed, for example, that my Sigma 17-70 f/2.8-4.5 zoom seems to have given better images on my 10Mp K10D than it seems to on my 14.6Mp K-7 (Pentax).

In view of the high resolution of modern cameras, such as the Canon 7D at 18Mp, is this going to become an issue, or am I simple mistaken?

If it is an issue, how do we tell which lenses are best suited to match our cameras?

[beejaybee]

is this going to become an issue,

It has been for some time. Very high pixel counts only "work" when very large amounts of sharpening are applied.

Sorry but the "fine detail" shown by most high pixel count cameras is largely processing artifact.

[Monobod]

is this going to become an issue,

It has been for some time. Very high pixel counts only "work" when very large amounts of sharpening are applied.

Sorry but the "fine detail" shown by most high pixel count cameras is largely processing artifact.

I find this to be a bit difficult to believe. I have some expensive prime lenses made by Pentax and these give enormous amounts of detail without the need for large amounts of sharpening. I think it is more to do with the resolving power of the lenses. If not, why would the camera makers be increasing the pixel counts? If the resolution goes down as the pixels increase, this would be detrimental to sales in the end. The image quality also depends on the sensor size with respect to high ISO noise of course.

I really think that it is the budget lenses that are the problem. The manufcturers are producing better quality, but more expensive lenses to overcome this shortfall. My question remains 'What can we do to relate specific lenses to camera sensors in terms of compatibility'? Do we just rely on test reports and reviews? Not many seem to address this issue directly at the moment.

[LargeFormat]

It is an issue from all accounts. I factored in CZ lenses when I bought my a900 for this reason. All the reviews said one needed quality glass and I rather followed them, and Roger's recommendations.

ISO100 film in a 35mm frame is estimated to be equivalent to 20MP which common sense would tell you needed a fair lens to resolve all the detail. Bearing in mind that the digital sensor doesn't work over its whole area as there are gaps between the sites it is more demanding on lenses. If you drop down to high resolution APSc cameras such as the 7D the demand on the lens is higher as you are trying to resolve ever smaller pixels. As for compacts ....

[daft_biker]

Sorry but the "fine detail" shown by most high pixel count cameras is largely processing artifact.

Indeed but one could argue that the fine detail in any photograph is "processing artifact".

Printing artifacts make good test subjects though. I hope Damo will forgive me for using a test subject we would all have close to hand



With the 40D and 50mm f/1.4 (at f/6.3) on a tripod I moved in until it was almost resolving the blocks of ink then took the same shot with the 7D. I think the 7D has done a better job of almost resolving the detail even if it was a setup intended to show that. The above image is 100% crops put side by side in PS.

[TimF]

is this going to become an issue,

It has been for some time. Very high pixel counts only "work" when very large amounts of sharpening are applied.

Sorry but the "fine detail" shown by most high pixel count cameras is largely processing artifact.

Nonsense.

It is true that the higher the pixel count the better the lenses are required for best results. It's also true that a higher pixel count has an impact on how much one can stop down before diffraction effects become evident - ie less than is true with a smaller pixel count (on a similar size sensor). Some of the very finest glass is diffraction limited at around f/5.6.

The good news is that the 12 megapixel or so mark is still perfectly satisfactory for most people's needs. A higher pixel count will render the finest detail better, but to double the resolution of a 12MP camera would require going to 48 megapixel.

[daft_biker]

Some of the very finest glass is diffraction limited at around f/5.6.

At what magnification and on what sensor?

[TimF]

Try here

[Meredith]

According to Leica they will have start to have problems with sensors outresolving lenses when the pitch between photosites on digital sensors drops below 5 microns.

[AlexMonro]

I believe some recent Canons (7D, 550D) have already got below that...

[daft_biker]

I believe some recent Canons (7D, 550D) have already got below that...

Ah but then you get to use Canon glass that can be good enough to be diffraction limited by f/4 never mind f/5.6

f/2.8


f/4


f/5.6

[TimF]

Ah but then you get to use Canon glass that can be good enough to be diffraction limited by f/4 never mind f/5.6

So, inferior then.

[daft_biker]

Ah but then you get to use Canon glass that can be good enough to be diffraction limited by f/4 never mind f/5.6

So, inferior then.

Nope, the diffraction limit gets lower as the pixel density goes up. I just neglected to mention anything about the image magnification or effective aperture either

[TimF]

Huh?! I have a feeling we may be speaking at cross purposes here. A diffraction limited lens is one in which the optical quality is as close to perfect as possible at maximum aperture, and is limited only by the effects of diffraction. The link in my 23:49 post on the 18th shows that with a digital LF Rodenstock lens. Others would include super telephotos from the likes of Canon etc.

This page shows the effect nicely when comparing pixel densities in the section headed "Visual Example: Aperture vs. Pixel Size" (both individual camera models and apertures are selectable).

A few years ago BJP ran a piece about a photographer using medium format digital. His back was 22 megapixel, but he wasn't interested in upgrading to one with say 39MP, saying that more megapixels had an impact on depth of field. That confused me back then, but it's obvious now that diffraction limitation is what he was on about in that stopping down too far causes a decline in IQ. That happens with any camera, film or digital, but occurs sooner the higher the pixel count on a given sensor size.

[OneTen]

I've just been reading about this and trying to get my head around it, probably not the best time as I'm full of a cold. From what I can gather it's all to do with Fairy Disks.

[Malcolm_Stewart]

... From what I can gather it's all to do with Fairy Disks.

Try (F)airy Disk for an explanation...

[TimF]

And some easy bedtime reading. There'll be questions afterwards!

[daft_biker]

There was a reason I asked about the magnification that was diffraction limited at f/5.6 as at low magnifications the pixel density is likely to be the limiting factor but as magnification goes up the limit will come down.

I'd have prefered to see the test in your first link done at infinity to show it was diffraction limited at the lowest magnification. (the result is likely to be the same but it would have been a tougher test IMO)

The article starts off:

Every photographer wants both maximum resolution and maximum depth of field. But unfortunately these two demands can be mutually exclusive.

Or as I think of it....
Every macro photographer wants maximum magnification, maximum resolution and maximum depth of field. But unfortunately these three demands can be mutually exclusive.

Macro photographers have been finding the limits of what lenses can resolve for years. It's not a new concept for some but in lower magnification shots stopping down is certainly an increasinly more difficult decision on higher density sensors if you want to be able to go pixel peeping.

Not too sure I'm that bothered about pixel peeping landscapes though....if it's good enough for A4 it's good enough for anything I've had printed

[TimF]

as magnification goes up the limit will come down.

I don't get where you're coming from on this. Diffraction on the sensor begins when the Airy discs begin overlapping by a certain amount. What difference would shooting at 5:1 show against 1:5? If you have a link or two to explain this it would be appreciated (and proper - though easy for laypeople to comprehend - scientific ones, not some forum guru's opinions please!

I just found this, which sets things out very well I think. A higher pixel count will render detail better up to a given aperture (he mentions f/64 in passing) when diffraction effects will mean there is no advantage over a lower megapixel count. That said, to achieve such limits in reality means shooting the very finest glass; if your lenses are incapable of resolving fine detail you won't get any benefit from using a higher megapixel count body.

[Overread]

as magnification goes up the limit will come down.

I don't get where you're coming from on this. Diffraction on the sensor begins when the Airy discs begin overlapping by a certain amount. What difference would shooting at 5:1 show against 1:5? If you have a link or two to explain this it would be appreciated (and proper - though easy for laypeople to comprehend - scientific ones, not some forum guru's opinions please!

Diffraction test

The title of each image shows the aperture used and the magnification of the shot and the set description outlines the testing method; each image is also uploaded at fullsize so you can really pixel peep as much as you want. There might be something else going on besides diffraction, but as far as I know this is showing that as magnification factor increases the diffraction limit moves.