Is resolution "lens limited"?

[Nemo]

Michael Reichmann wrote this in his comment about the new Canon 1Ds Mark III:

My main concern is that the 16MP 1Ds MKII was in many cases lens limited. Only the best lenses, and these mostly "L" lenses in the telephoto range, were up to the task. Though the jump from 16MP to 21MP is modest, and certainly worthwhile, it concerns me that the new MKIIIs is going to be even more lens-limited than the previous generation.

I don't think actual cameras are lens limited at all. The resolution of the 1Ds Mark III is just the same of the Canon 20D or 30D, because the pixel spacing is 6,4 microns (the Mark II was 7,2 microns), this is, 80 lp/mm of resolving power (the Mark II was 70 lp/mm). The sensors of the 400D or 40D have more resolving power yet.

However, actual lenses' resolution is even higher. The contrast drops when the resolved detail increases, but you don't need a 50% of contrast in order to see it. On the other hand, the resolving power of a lens on axis is much higher than the resolving power on the field, or the corners. The vignetting, chromatic aberrations, filters-induced astigmatism, etc. are a different story.

There are many factors affecting the sensor’s ability in recording detail, but we will simplify the explanation suggesting a simple rule. The real detail finally captured by the camera system (sensor and lens) can be approximated from this formula:

R = 1 / (1/r1 + 1/r2)

where R is the total resolution of the camera system (in line pairs per millimeters), r1 is the resolution provided by the lens and r2 is the resolution of the sensor. Only when one of the “r” tends to infinity the total resolution of the system, R, tends to the other “r”. This means that the total resolution of the sensor or the lens never is fully brought into the photograph. Actual sensors for reflex and rangefinder cameras typically have resolutions from 60 to 100 lp/mm, but good lenses show much higher numbers (at least on axis). The greater the resolution of the lens is, the better we will be exploiting the resolution of the sensor.

http://www.luminous-landscape.com/essays/L...rspective.shtml

R.

[michael]

It isn't about resolution. It's all of the other factors that you mention. A higher resolution sensor allows larger prints (or greater cropping) and this means that all of the lens's flaws are exposed to a greater extent.

The fact that Canon is (slowly) updating its lens lens (particularly at the medium and wide ends of the range) shows that they are aware of this issue.

Michael

[madmanchan]

Agreed. As I recall, Canon did the same thing when the original 1Ds came out. The mid-range L zoom of the time, the 28-70 f/2.8L, was replaced by the 24-70 f/2.8L.

[Graeme Nattress]

Given that the 1DsMKIII has the same pixel pitch as the 20D, and you can get the glass to outresolve the 20D, we're probably fine in the centre of the image. However, I would be concerned about what is going on at the edges of the full frame sensor, and how the lenses cope there.

[BJL]

I would not worry about the 1DsMkII failing to provide improved resolution when used with appropriate lenses, meaning good primes and excellent zooms, since surely that is what one would use if seeking even more resolution that the 1DsMkII provides and paying $8,000 for the opportunity.

But then again, the resolution increase will only be about a modest one eighth (12.5%), since that is the increase in linear pixel density. That seems mostly in the realm of "slightly more cropping latitude while still meeting my client's rigid requirement of 300ppi uninterpolated for a double page spread". (At 300ppi, the change is from 11.1"x16.64" to 12.48"x18.72".)

Even with lower pixel count DSLRs and film of less than the maximum available resolution, there are noticeable differences in sharpness and resolution between different lenses, so to some extent, resolution is already "partially lens limited" (except perhaps with the sharpest available lens.) And yet when one upgrades from those cameras to a higher resolution film or sensor, overall resolution is seen to improve with many lenses. This is because, as has been mentioned, even when the lens partly limits resolution, so does the "sensor", so improving either with improve the overall result.

Sensor resolution improvements would only fail to improve overall resolution if the lens limits were dominant, so that very little resolution could be gained by improving the sensor. I do not believe that the 1DsMkII is near that point, at least with good primes and excellent zoom lenses, though the wide angles seem to be closest to this problem. And look at some lenses that Canon has offered recently: new versions of the 14/2.8 and 16-35/2.8L.

[Nemo]

Erwin Puts wrote a very good article titled "the truth about digital lenses":

http://www.imx.nl/photo/optics_2/11_the_tr...t_digital_.html

I would recommend this one too: "Canon digital and 35mm format issues".

http://www.imx.nl/photo/viewpoint/text_5.html

Geoffrey Crawley writes a very good article in the actual issue of Amateur Photographer, about lens design.

To sum-up, digital lenses need

1) different coatings,

2) more control of spherical aberration (this aberration explain the famous "focus shift"),

3) redefinition of the focus plane and tighter tolerances (the film had 10-20 microns of thickness, but the sensor hasn't... although it has filters on it).

... but the software corrections of problems like vignetting, distortion and fringes make the life of the optical designer a bit easier. You can undercorrect several aberrations and get additional degrees of freedom for overcorrection of the other aberrations. Therefore, "digital" lenses can be  designed under a much diferent set of parameters and priorities.

Canon isn't redesigning the lenses from scratch. They are making corrections to the basic (unchanged) design. In addition, optical progress in lens design doesn't stop, and minor improvements are welcome in any case (even for film cameras).

I would like to write about the new Leica Summarit-M lenses, for instance. These lenses will be top performers, even if the costs are reduced!

[Nemo]

It isn't about resolution. It's all of the other factors that you mention. A higher resolution sensor allows larger prints (or greater cropping) and this means that all of the lens's flaws are exposed to a greater extent.

The fact that Canon is (slowly) updating its lens lens (particularly at the medium and wide ends of the range) shows that they are aware of this issue.

Michael
[a href=\"index.php?act=findpost&pid=134475\"][{POST_SNAPBACK}][/a]

Thank you for your kind response.

I agree. I thought this issue needed a bit of clarification. The higher the resolution of the sensor is, the more resolving power of the other device (lens) is exploited. The flaws can be more exposed, because you will have a gigant file (21 megapixels) and the information really recorded (for instance, detail at high levels of contrast) could not justify the file size.

Best.

R.

[Nemo]

Even with lower pixel count DSLRs and film of less than the maximum available resolution, there are noticeable differences in sharpness and resolution between different lenses, so to some extent, resolution is already "partially lens limited" (except perhaps with the sharpest available lens.) And yet when one upgrades from those cameras to a higher resolution film or sensor, overall resolution is seen to improve with many lenses. This is because, as has been mentioned, even when the lens partly limits resolution, so does the "sensor", so improving either with improve the overall result.

Sensor resolution improvements would only fail to improve overall resolution if the lens limits were dominant, so that very little resolution could be gained by improving the sensor. I do not believe that the 1DsMkII is near that point, at least with good primes and excellent zoom lenses, though the wide angles seem to be closest to this problem. And look at some lenses that Canon has offered recently: new versions of the 14/2.8 and 16-35/2.8L.
[a href=\"index.php?act=findpost&pid=134619\"][{POST_SNAPBACK}][/a]

Your explanation is clearer than mine. Thank you very much!

R.

[Nemo]

Given that the 1DsMKIII has the same pixel pitch as the 20D, and you can get the glass to outresolve the 20D, we're probably fine in the centre of the image. However, I would be concerned about what is going on at the edges of the full frame sensor, and how the lenses cope there.
[a href=\"index.php?act=findpost&pid=134503\"][{POST_SNAPBACK}][/a]

Lenses have less resolving power at the edges of the image circle. The only problem with higher resolution sensors in this area (the field and corners) is this: you exploit the resolving power more in the axis (center) and less in the field.

The digital image is a matrix of pixels, and it is like a bottle. You can fill it more or less. The low-pass filter is like a stone into that bottle. You cannot fill up the bottle. The Bayer mosaic is another stone into the bottle. Less liquid (information) can be spilt into. The lens' MTF is another limit, because lenses cannot transmit the finer detail at 100% contrast. Etcetera.

We are getting more liquid increasing the size of the bottle, this is, increasing the number of pixels. I expect to see very soon new developments aimed at increasing the system eficiency! It would allow to get the same information (liquid) in smaller bottles. For instance, Foveon-class sensors, better lenses, better filters...

[MarkKay]

I tried 2 different 16-35 II and compared to my version I.  The first version II lens was the better at edge resolution and distortion at 16mm-24mm range from   2.8 to 5.6 compared to  my version I.  At f8, very close.  At 35mm, the newer lens was not as good as my version I.  The second lens I tested with rare exception was  generally worse than my version I lens.  In general, the version II lens had less CA.  I kept my version I lens. So i hope the improved 14mm is more robust than the differences I observed beween the 16-35 lenses.

I would not worry about the 1DsMkII failing to provide improved resolution when used with appropriate lenses, meaning good primes and excellent zooms, since surely that is what one would use if seeking even more resolution that the 1DsMkII provides and paying $8,000 for the opportunity.

But then again, the resolution increase will only be about a modest one eighth (12.5%), since that is the increase in linear pixel density. That seems mostly in the realm of "slightly more cropping latitude while still meeting my client's rigid requirement of 300ppi uninterpolated for a double page spread". (At 300ppi, the change is from 11.1"x16.64" to 12.48"x18.72".)

Even with lower pixel count DSLRs and film of less than the maximum available resolution, there are noticeable differences in sharpness and resolution between different lenses, so to some extent, resolution is already "partially lens limited" (except perhaps with the sharpest available lens.) And yet when one upgrades from those cameras to a higher resolution film or sensor, overall resolution is seen to improve with many lenses. This is because, as has been mentioned, even when the lens partly limits resolution, so does the "sensor", so improving either with improve the overall result.

Sensor resolution improvements would only fail to improve overall resolution if the lens limits were dominant, so that very little resolution could be gained by improving the sensor. I do not believe that the 1DsMkII is near that point, at least with good primes and excellent zoom lenses, though the wide angles seem to be closest to this problem. And look at some lenses that Canon has offered recently: new versions of the 14/2.8 and 16-35/2.8L.
[a href=\"index.php?act=findpost&pid=134619\"][{POST_SNAPBACK}][/a]

[BJL]

Mark, I have no experience with the 16-35's, but you are not the first to suggest that the 16-35 Mk II does not improve resolution much. (At least Canon knows where work is needed in order to get the most out of the 1Ds series!)

Maybe the situation is that for wide angles in particular, high budget, high resolution photography is still best done with primes! Not a great handicap for the 1Ds when the competition is medium format, almost a "primes only" zone, and with no MF lens covering a FOV as wide as 16mm on a 1Ds.

So it will be interesting to see the performance of the 14/2.8 II, and whether Canon adds a few more high end wide angle primes.

[MarkKay]

I know another photographer who is well respected who had similar experiences. i should also point out that i tried 3 16-35mm I lenses to pick the one i currently have so some of the debate about the differences between version I and II may be due to the copy to copy variation that seems to plague the canon wide angle zooms.  In honesty, I had a tamron 14mm that I had received from Jack Flesher that was much better with edge sharpness, and about equal with distortion (but perhaps a little less good with CA)   than the canon 14mm prime and a fraction of the cost.  So again, I am going to be interested in seeing the differences in the version I vs II prime on a full frame high resolution Canon. I suspect or at least hope the copy variation differences with the prime lens will be less problematic so that the  interpretation based on everyones comparisons are more concordant.

Mark, I have no experience with the 16-35's, but you are not the first to suggest that the 16-35 Mk II does not improve resolution much. (At least Canon knows where work is needed in order to get the most out of the 1Ds series!)

Maybe the situation is that for wide angles in particular, high budget, high resolution photography is still best done with primes! Not a great handicap for the 1Ds when the competition is medium format, almost a "primes only" zone, and with no MF lens covering a FOV as wide as 16mm on a 1Ds.

So it will be interesting to see the performance of the 14/2.8 II, and whether Canon adds a few more high end wide angle primes.
[a href=\"index.php?act=findpost&pid=134874\"][{POST_SNAPBACK}][/a]

[Nemo]

I want to recommend a new article written by Erwin Puts. It is very, very interesting:

http://www.imx.nl/photo/technique/leica_le...digital_ne.html

 


R.

[bjanes]

I want to recommend a new article written by Erwin Puts. It is very, very interesting:

http://www.imx.nl/photo/technique/leica_le...digital_ne.html

 
R.
[a href=\"index.php?act=findpost&pid=136265\"][{POST_SNAPBACK}][/a]

The quote below is from Mr. Puts' essay:

"Handheld photography with medium speed films and moving objects (street scenes) is limited to at most 20 lp/mm and often less. The 40 lp/mm limit was established by Zeiss researchers as a practical limit for high definition 35mm photography. My own tests have indicated that is not easy to reach this limit in normal situations. A small focus mismatch or a camera vibration is enough to destroy any ambition for high definition images."

If you want to get all the resolution that your EOS 1DsM3 can produce, you had better mount it on a sturdy tripod and use mirror lockup in addition to using the best lenses. For hand held work you probably won't exceed the 40 lp/mm limit described by the Zeiss researchers.  

Here are the Nyquist limits for some common dSLRs. These cameras usually resolve at about 80% of Nyquist or less due to limitations of the Bayer sensor and demosaicing. Before spending a lot of money on ultra high resolution, you should think about how much resolution you can actually achieve under your field conditions.  Image stabilization may improve the upper limit, but I have no data on that.



Bill

[Ray]

The quote below is from Mr. Puts' essay:

"Handheld photography with medium speed films and moving objects (street scenes) is limited to at most 20 lp/mm and often less. The 40 lp/mm limit was established by Zeiss researchers as a practical limit for high definition 35mm photography. My own tests have indicated that is not easy to reach this limit in normal situations. A small focus mismatch or a camera vibration is enough to destroy any ambition for high definition images."

[a href=\"index.php?act=findpost&pid=136280\"][{POST_SNAPBACK}][/a]

Bill,
One can't disagree with that, but we've progressed a lot since the days of film. DSLRs have much better performance at high ISO than film ever had. Combine that performance with a modern IS lens then maximally sharp hand-held shots should be possible, although the shutter speed might still have to be faster than the 1/FL(35mm) rule would suggest, even with an IS lens.