Pages: [1] 2 3   Go Down

Author Topic: Now that sensors have stabilised, will lenses be the next area of focus?  (Read 23526 times)

dreed

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1715

Reading DxO's latest review of the Sigma 35mm/f1.4 on the Nikon D800, it would seem that Sigma have really hit the mark with what appears to be a very superb lens. Looking at this as yet another example of where 3rd party manufacturers are upping the ante in terms of lens quality, I'm wondering if this is really the beginning of a period during which we see a significant upturn in lens performance & quality, especially now that there are companies/websites such as DxO that are putting lenses under the microscope and their flaws being bared for all to see.

Are we thus past the point where sensor changes and improvements (at least so far as megapixels goes) are they key determining factor in resulting IQ and into a phase where it is improvements in lens quality that will be dominant?
Logged

RFPhotography

  • Guest

Have sensors stabilised?  It was only a year ago that we got the 36MP D800.  And now the new D7100. 
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365

I'd say, not. Even before Canon released its first DSLR over 11 years ago, the Swedish magazine, Photodo, published detailed MTF results of lenses alone, without their being attached to a camera body. Such results were an excellent guide for choosing a lens.

We've now regressed to the point where lens tests are never separated from the performance of a particular model of camera, except for the manufacturers' theoretical MTF charts which rarely go beyond 30 lp/mm.

I find the Photozone Imatest MTF 50 results quite useful, but they have warnings on their site against comparing lenses tested on different cameras, for the obvious reasons that different sensors have different pixel counts, different strengths of AA filter, and the processed RAW images may have been subjected to different qualities of processing according to the age and sophistication of the RAW converter used.

The lens tests at Photozone and DXO are not really lens tests but system tests which always include sensor performance.

If we want to 'up the ante' on lens development, I think we should start doing Photodo-style MTF tests again, but at higher resolutions, up to 80 or 100 lp/mm. As pixel density increases, there should never be a need for an AA filter and the pixel density alone should be a sufficient indication of the resolving power of a sensor.
Logged

RFPhotography

  • Guest


As pixel density increases, there should never be a need for an AA filter and the pixel density alone should be a sufficient indication of the resolving power of a sensor.


It will be interesting to see if the D7100 is any better in that regard to the D800e.  The D800e is still prone to moirĂ© and aliasing.
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913

As pixel density increases, there should never be a need for an AA filter and the pixel density alone should be a sufficient indication of the resolving power of a sensor.

Hi Ray,

That's correct, but it would require something like a sensel pitch of 1 - 1.1 micron to have diffraction dictate the resolution even at wide apertures. Also the different sampling densities for Red/Blue versus Green require very dense sampling to avoid all False Color artifacts. We're not quite there yet, and a good AA-filter is more effective than diffraction.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051

We've now regressed to the point where lens tests are never separated from the performance of a particular model of camera, except for the manufacturers' theoretical MTF charts which rarely go beyond 30 lp/mm.

I find the Photozone Imatest MTF 50 results quite useful, but they have warnings on their site against comparing lenses tested on different cameras, for the obvious reasons that different sensors have different pixel counts, different strengths of AA filter, and the processed RAW images may have been subjected to different qualities of processing according to the age and sophistication of the RAW converter used.
If you are interested in the MTF of a lense at really high spatial frequencies, you might want to use a test-chart of e.g. a slowly swept sinoid, meaning that you know what (single) frequency is fed into the system at a particular spot. If you then use something like the D800E (or the D7100) without a AA-filter, you know that the camera sensor has a wider frequency response (even to aliasing frequencies).

Presence of significant energy aliased onto lower frequencies might be used a robust measure of the lenses MTF at frequencies higher than a Nyquist-adhering sensor is normally able to peer into.

Poor fill-factor might give even better measurements, but I think the combination of poor fill factor, high sensel density and no OLPF is  unheard of?

-h
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

I just looked at two lenses on two cameras, a Minolta 100/2.8 macro and a Sony 24-70/2.8ZA , on 24 MP APS-C (Sony SLT 77) and on a 24 MP FF (Sony SLT 99). Both lenses at f/5.6 on axis manually focused with live view. The image were processed in LR 4.3 with no sharpening and analyzed for MTF using Imatest. Plotting was MTF vs. lp/mm.

The better of the two lenses (centrally) was the 24-70/2.8 ZA. Nyquist is at around 125 lp/mm on A77 and at 83 lp/mm on the A99.

Plots below.

Best regards
Erik


Reading DxO's latest review of the Sigma 35mm/f1.4 on the Nikon D800, it would seem that Sigma have really hit the mark with what appears to be a very superb lens. Looking at this as yet another example of where 3rd party manufacturers are upping the ante in terms of lens quality, I'm wondering if this is really the beginning of a period during which we see a significant upturn in lens performance & quality, especially now that there are companies/websites such as DxO that are putting lenses under the microscope and their flaws being bared for all to see.

Are we thus past the point where sensor changes and improvements (at least so far as megapixels goes) are they key determining factor in resulting IQ and into a phase where it is improvements in lens quality that will be dominant?
« Last Edit: March 19, 2013, 05:41:38 pm by ErikKaffehr »
Logged
Erik Kaffehr
 

Rhossydd

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3369
    • http://www.paulholman.com

especially now that there are companies/websites such as DxO that are putting lenses under the microscope and their flaws being bared for all to see.
It's nothing new. UK photo magazines were publishing MTF charts in lens reviews back in the 1970s.
If you really cared about lens performance back then you bought from RG Lewis in High Holborn who individually MTF tested every lens they sold and supplied the chart with the lens. Sadly they're now long gone and I suspect were the only photo dealer anywhere to offer such a service.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365

Hi Ray,

That's correct, but it would require something like a sensel pitch of 1 - 1.1 micron to have diffraction dictate the resolution even at wide apertures. Also the different sampling densities for Red/Blue versus Green require very dense sampling to avoid all False Color artifacts. We're not quite there yet, and a good AA-filter is more effective than diffraction.

Cheers,
Bart

Hi Bart,
Clearly the problem of aliasing is on a sliding scale, as with most things. The higher the pixel density, the less significant the problem of aliasing becomes. It may not completely disappear in all circumstances until the pixel pitch on the sensor has become as small as 1 micron, as you claim, but surely only those who take photographs for scientific purposes would be concerned about such perfection.

I can't help recalling in this context my comparisons of the Canon 40D and 50D when photographing the Australian $50 bill, using LiveView, tripod and the same Canon 50/1.4 lens on each camera. Although the 50D, with a 50% greater pixel count, provided more detail at all apertures down to and including F16, the most glaring differences between the images, particularly at F5.6 and F8, were the very obvious aliasing and color moire in the 40D shots, and the relative absence of it in the 50D shots.

Both cameras have an AA filter, thus demonstrating the principle that having an AA filter in front of the sensor does not guarantee the complete avoidance of all aliasing artifacts. To do that, the AA filter would have to be so strong that customers would complain about the soft images the camera produced.

It would be interesting to repeat the experiment with the $50 bill and include the D7100 in the test. My guess is that the D7100 would produce noticeably more detailed results than the 50D, but without any increase in aliasing artifacts compared with the 50D, which produced only minor aliasing on this particular target. We would then have a situation whereby a camera without an AA filter produces less aliasing than another model, the Canon 40D, which does have an AA filter.

Cheers!
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365

If you are interested in the MTF of a lense at really high spatial frequencies, you might want to use a test-chart of e.g. a slowly swept sinoid, meaning that you know what (single) frequency is fed into the system at a particular spot. If you then use something like the D800E (or the D7100) without a AA-filter, you know that the camera sensor has a wider frequency response (even to aliasing frequencies).

Hi h,
I'm mainly interested in the MTF results of lenses for purchasing decisions. Those Photodo MTF results were a great buying guide. If I already own and use the lenses, I know their strengths and weaknesses.
My main concern when I buy a new lens is that I can expect it to be better than what I already have, at least in some respects.

For example, ever since buying a Nikon D7000 about 18 months ago I've been searching for a telephoto zoom with Nikon mount and VR (OS or VC) that at least matches the performance of my Canon 100-400. The three main contenders were the old Nikkor 80-400 VR, the Sigma 150-500 OS and the latest version of the Sigma 55-500 which now has OS.

I find it somewhat ridiculous that in this modern age of sophisticated technological development I can't find any reliable MTF results that compare all four lenses, including the Canon 100-400 which is my benchmark.
I can eliminate the old Nikkor 80-400 because there's so much anecdotal evidence that the Canon is the better lens. Both lenses have been around for a long time.
I can eliminate the Sigma 150-500 OS, but with less confidence because there are no comparisons at either Photozone or DXO. That leaves the Sigma 55-500 OS which, anecdotally, seems to be more highly regarded than the 150-500 and is also more expensive than the 150-500, which tends to imply it should be sharper.

Of course, now that Nikon has upgraded its 80-400, that would seem to be the logical choice for me. However, the Sigma 55-500 is about $1,000 cheaper and has a more useful range of focal lengths. I'd really like to know, for example, whether the new Nikkor at 400mm can produce the same or better detail than the Sigma at 500mm. If it can, and is also sharper than the Sigma at other focal lengths down to 80mm, then the matter is settled. I'd be prepared to pay the extra $1,000 and save a few hundred grams in weight, which is also a consideration for me.

Cheers!
Logged

Corvus

  • Newbie
  • *
  • Offline Offline
  • Posts: 28

now that there are companies/websites such as DxO that are putting lenses under the microscope and their flaws being bared for all to see.

If you have to put a lens under the "microscope" before you can "see" the flaws what real world practical relevance does such a test have for most photographers?
Perhaps any difference that makes no real world practical difference, at least to most photographers, is no difference.

Here's a novel idea - snap on a lens take some pictures and see if you are satisfied with the results.

Logged

dreed

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1715

If you have to put a lens under the "microscope" before you can "see" the flaws what real world practical relevance does such a test have for most photographers?
Perhaps any difference that makes no real world practical difference, at least to most photographers, is no difference.

If all lenses were the same and they all resolved at say 1lpmm and vignetted the same then your level of expectation would be that and you would not think that anything else could be better.

But that's not the case as there are many different lenses.

So what lens testing does is give us a quantifiable and objective way to explain what we can often see with our eyes but only use subjective words to describe. Quite possibly past a certain point it becomes hard to pick the difference with your eyes except that with digital photography and 1 to 1 display of pixels, what was not visible becomes visible. This then forms the foundation for knowing that what we've got could be better...
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

Nice to know the flaws before you buy the lens. It may also take a long time until flaws are found.

Best regards
Erik

If you have to put a lens under the "microscope" before you can "see" the flaws what real world practical relevance does such a test have for most photographers?
Perhaps any difference that makes no real world practical difference, at least to most photographers, is no difference.

Here's a novel idea - snap on a lens take some pictures and see if you are satisfied with the results.


Logged
Erik Kaffehr
 

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/

Maybe.

Considering the rumored performance of the new Nikon 800mm f5.6, Zeiss 55mm f1.4,... it feels like some breakthroughs are being made in lens design when cost is not an issue.

Now, Leica's latest R lenses, like the 180 f2.8 APO, were already at that level of performance almost 10 years ago. So it may just be that companies not interested in the past in the very high end segment are simply now trying to address it also.

Simulation software and computational resources have made huge progress these past few years and those are probably the main reasons for the progress we can now witness. Complex optimization problems can now be solved and large virtual test matrix can be run realistically.

The Sigma 35mm f1.4 is in my view a breakthrough from a cost standpoint. Its performance is remarkable, but in the end pretty similar to that of the Nikkor 35mm f1.4. As far as I am concerned I'll wait a few more months to see whether Nikon releases a more compact 35mm f1.8. My main usage for these lenses is street shooting and I find both the Nikkor and the Sigma just too large/heavy.

Cheers,
Bernard

Rob C

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 24074

No disrespect, but I sometimes wonder if any of this has a heap to do with photography in the sense of pictures.

I look at a helluva lot of pictures on the web and in various newspapers and magazines that lie around Spanish bars. Just yesterday afternoon I was having a coffee and reading a fairly thick fashion magazine called Woman-something, and though the images were all very nice, it didn't escape me that despite being totally up-to-date, the photographic styles were hardly changed from the 70s; in fact, one of the fashion features was all about current references to the period. To be brutally honest, I don't think any of it has changed very much since the 70s/80s other than that the new kids on the block seem perfectly happy to show more tit and then disguise it as further plastic to match the faces. A tit doth not a fashion image make. And it's deception to pretend that it does. I'm no monk, but I have yet to see women sitting in bars and restaurants with a tit on display. Or even a matched pair. So the big stride ahead comes down to Photoshop and the loss of skin textures. (Making me question again the use of very young models: if you are going to PS everything, what's the point of starting off with younger skin?)

But cameras? Lenses? From the evidence of the printed work, we might as well still be using TXP or Kodachrome.

All this stuff is so damned academic that it's almost a joke. Where do I get to see the photographs that show the value of all this technical jargon and manufacturing 'progress' that's being bandied about? I see very few impressive new images but reams and reams of crap about bokeh, various curves and that's about it. The new photography, then, the beneficiary of all this stuff? The thrill is in the measurements? It's in the buying of something theoretically 'better' than what one already owns? It sure ain't in the printed pix.

Rob C

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913

All this stuff is so damned academic that it's almost a joke. Where do I get to see the photographs that show the value of all this technical jargon and manufacturing 'progress' that's being bandied about? I see very few impressive new images but reams and reams of crap about bokeh, various curves and that's about it. The new photography, then, the beneficiary of all this stuff? The thrill is in the measurements? It's in the buying of something theoretically 'better' than what one already owns? It sure ain't in the printed pix.

Hi Rob,

Maybe the print quality requirements are low enough to not challenge the input quality?
Did you ever compare modern large format output quality with some of the stuff from yonder years, and the relative ease to achieve it.

We no longer need something like this to produce posters, do we?

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Corvus

  • Newbie
  • *
  • Offline Offline
  • Posts: 28

you would not think that anything else could be better.

I think the use of the relative term "better", rather than an absolute term, goes to the heart of the problem.

I have a modest little 300 buck 35mm lens that gives me excellent optical performance. So much so that I would be satisfied if all my lenses, of whatever focal length gave me comparable optical performance. At some absolute point of optical performance enough is enough. My pictures may have problems but they are not problems that will be solved by the latest $2000 super lens. Is a car that can cruise all day at 160 miles per hour twice as good as a car that can cruise all day at 80 mph - given the real world constraints of how we actually use a car?

On the other hand - I have a camera that now gives me excellent results at an ISO of 1600. Give me the same identical camera but with a high ISO performance at, say, an ISO of 12000 that I now get at 1600 and I would buy it in a heartbeat. The absolute level of sensor performance is still not good enough for my purposes while at least some of my lens' are.

Speaking only for myself - decent glass is critical but not all that hard to obtain while there are much more important practical needs that are still not being met.
« Last Edit: March 22, 2013, 06:53:22 am by Corvus »
Logged

Corvus

  • Newbie
  • *
  • Offline Offline
  • Posts: 28

I see very few impressive new images but reams and reams of crap about bokeh, various curves and that's about it. The new photography, then, the beneficiary of all this stuff? The thrill is in the measurements? It's in the buying of something theoretically 'better' than what one already owns? It sure ain't in the printed pix.

Yep
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/

All this stuff is so damned academic that it's almost a joke. Where do I get to see the photographs that show the value of all this technical jargon and manufacturing 'progress' that's being bandied about? I see very few impressive new images but reams and reams of crap about bokeh, various curves and that's about it. The new photography, then, the beneficiary of all this stuff? The thrill is in the measurements? It's in the buying of something theoretically 'better' than what one already owns? It sure ain't in the printed pix.

Agreed overall, but how about large prints?

Cheers,
Bernard

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051

If you have to put a lens under the "microscope" before you can "see" the flaws what real world practical relevance does such a test have for most photographers?
Perhaps any difference that makes no real world practical difference, at least to most photographers, is no difference.

Here's a novel idea - snap on a lens take some pictures and see if you are satisfied with the results.
I expect my lenses to last 10 years +. I expect my cameras to "last" less than 5 years. Therefore it is relevant to have some insight into how any given lens will perform on the kind of cameras that is on the market 5-10 years from now.

-h
Logged
Pages: [1] 2 3   Go Up