1DS Mark III - Lens Resolving Concerns

As far as I understand, the resolving power of current L series lenses are not enough to cope with the increased resolution of the new 1DS Mark III. There has been talk for a long time that the resolving power of current L lenses already struggles to keep up with the 1DS Mark II sensor and I myself have felt that way at times.

At one time there was speculation that the next generation 1DS after the Mark II, when Canon makes the jump to 20+ MP, would be a hybrid of something between a 35mm camera and a medium format body with a whole new lens system designed specially to provide more resolving power for a super higher resolution sensor.

Since Canon has not gone that route, I am wondering if they plan to offer another solution to address the issue of current lenses which won’t be able to keep up with a sensor which produces images at over 20MP.

Thoughts?

Marc Schultz said:

As far as I understand, the resolving power of current L series lenses are not enough to cope with the increased resolution of the new 1DS Mark III. There has been talk for a long time that the resolving power of current L lenses already struggles to keep up with the 1DS Mark II sensor and I myself have felt that way at times.

Click to expand...

Welcome to OPF, Mark.

There are several things going on, but lens resolution in general is more than adequate for the new 1Ds Mark III. The complaints about lenses you've read about have more to do with corner performance of wide angle lenses. With a slightly denser sampling (smaller sensels) of those corners the aberrations will just be more accurately recorded .

Many lenses have a resolution that exceeds 150 lp/mm in large parts of the image circle. The sensor can 'only' resolve 78 lp/mm at best (78 cycles/mm = Nyquist limit for a 6.4 micron sampling density), and then there is an AA-filter that reduces the resolution of the lens projection as well. Lens resolution is not the issue, residual lens aberrations may be an issue, depending on shooting conditions.

At one time there was speculation that the next generation 1DS after the Mark II, when Canon makes the jump to 20+ MP, would be a hybrid of something between a 35mm camera and a medium format body with a whole new lens system designed specially to provide more resolving power for a super higher resolution sensor.

Click to expand...

Depending on the speculation, the real concern is that the Dynamic Range will be adversely impacted by more and thus smaller sensels. Yes, we are at a dividing point between what's physically possible between resolution and dynamic range within a given 24x36mm area. Increasing the active sensel surface will reduce resolution, and increasing resolution will reduce dynamic range. The only way one can break out of that trade-off and have both higher resolution and dynamic range, is by increasing the physical size of the sensor array, which in turn requires lenses that cover a larger image circle and thus requires a different type of camera with different lenses.

Since Canon has not gone that route, I am wondering if they plan to offer another solution to address the issue of current lenses which won’t be able to keep up with a sensor which produces images at over 20MP.

Thoughts?

Click to expand...

I hope it is clearer now that the lenses are not holding the sensor back, they resolve twice the amount of detail that the sensor array could in large parts of the image, but there are design issues that lead to relatively poor quality areas compared to the center of the image circle, and that difference in the same image is sometimes disturbing. We all would like better lenses, but that's a different issue, and it involves different prices as well.

Bart

That is all very interesting Bart and thanks for sharing information that clears up the mystery of 150 lp/mm being enough resolving power or not for that many pixels within a sensor area of just 24x36mm.

The active sensel surface you are referring to is what I think I refer to as “sensor receptor sites”, which is also known as the pixel pitch measured in microns.

It used to be that people would talk about a minimum pixel pitch of 9 microns in order for a sensor to produce low noise, high quality pixels.

But then the envelope changed when Canon started putting out cameras like the 1DS Mark II which has a pitch of about 7.2 microns, below the 9 micron threshold, and which produces images that are lower noise for example than the 1DS which is a nosier camera, but does employ the perceived optimum pixel pitch of 9 microns.

So at that point people started redefining the threshold for optimum pixel pitch, mainly because of the improved technology of DSLR processors being able to control noise at smaller pixel pitches. Some have even predicted that one day we will be able to turn up the frequency high enough on a sensor which has only a 5 micron pitch and still produce smooth, colorful, noise free pixels.

As the 1DS Mark III has a pixel pitch of just 6.4 and claims to have offer less noise than the 1DS Mark II, it means the 5 micron pixel pitch envelope may be getting closer to becoming a reality very soon.

So if the lenses don’t have a resolving issue, but only a light softness around the edges, and the newer processors can handle the noise from smaller, higher charged receptor sites and smaller pixel pitched sensors, then I don’t see much to be worried about anymore.

In my case however, already having a 1DS Mark II, it makes more sense for me to buy a Phase One P25, which can now be had second hand for close to the same money as a 1DS Mark III. Although both will offers the same resolution, the sensor on the Phase One delivers a 22MP image from a sensor with a 9 micron pixel pitch and allows me to employ a Hasselblad with Zeiss glass.

The other problem for the first 2 years that the 1DS Mark III will face will be convincing the commercial market that a 35mm camera at any resolution will be able to produce an image as high quality as a medium format digital back on a Hasselblad. So if the clients are not going to buy it, at least right away, then why should I is really the big question I guess.

Cheers…

Marc Schultz said:

The other problem for the first 2 years that the 1DS Mark III will face will be convincing the commercial market that a 35mm camera at any resolution will be able to produce an image as high quality as a medium format digital back on a Hasselblad. So if the clients are not going to buy it, at least right away, then why should I is really the big question I guess.

Click to expand...

Hi Marc,

A huge welcome to OPF. Glad you stepped in on such an important topic. We look forward to a lot more of this high level participation and maybe sharing your images too

As far as the 1DsIII is concerned, it has no competition to speak of for what the Canon flagship excels: the ability to use fast autofocus lenses such as the 50 1.2L, the 85 1.2L, the 70-200 IS and the 28-300 L and so forth.

For sure, a the wide end, the digital backs are supreme. For architecture and art reproduction the MF backs have no competition except for LF film. For fashion the digital backs are again masters, but as has been demonstrated by Benjamin Kanarek here even the Pentax K10D and the Cano Rebel XTi can do totally professional fashion that is competitive!

So I don't think Pro Photographers are going to need much convincing either way. Compared to film, the best MF backs are not that expensive. However the best 35mm format DSLR's (even with 1.2,1.3 or 1.6) multipliers all have their own advantages.

It ultimately depends on your shooting style and needs of your clients and those decisions have in large part already already been made. I can't see that the 1DsIII really alters the balance either way.

Asher

Hey !
just read European CPS website…

Down below that page

New lens

Today’s launch coincides with the release of the EF 14mm f/2.8L II USM: Canon’s new professional L-series ultra wide-angle lens for the photographer requiring a slightly different perspective. As with all L-series lenses and the Speedlite 580EX II, the EF 14mm f/2.8L II USM incorporates environmental sealing to match that of the EOS-1Ds Mark III, creating a fully weatherproof camera system. Aspherical and UD lens elements produce exceptional image quality with improved corner-to-corner sharpness. The ring-type USM delivers fast, silent auto focus with full-time manual focus override. A circular aperture produces pleasing bokeh at wide apertures and Super Spectra coatings suppress ghosting and flare by suppressing internal reflections from lens element and sensor surfaces.

Click to expand...

Great news isn't it?

Nicolas Claris said:

Great news isn't it?

Click to expand...

Yes, although it's IMHO a specialist's lens, it's a great addition to the line-up. If the generic MTF curves on the Japanese website can be taken as an indication, the lens has a stellar performance when stopped down, all the way to the (FF) corners.

Bart

Hi, Bart,

Bart_van_der_Wolf said:

The sensor can 'only' resolve 78 lp/mm at best (78 cycles/mm = Nyquist limit for a 6.4 micron sampling density)

Click to expand...

It is also possible that Kell factor considerations might make the "theoretical" resolution limit even lower than that.

Nicolas, will the event of a new 14mm, put the Sigma 12-24 in the shade? I guess the Sigma will still have the edge on low distortion (and of course price)

Doug_Kerr said:

It is also possible that Kell factor considerations might make the "theoretical" resolution limit even lower than that.

Click to expand...

Hi Doug,

I agree, the Nyquist frequency mentioned is the absolute highest spatial frequency that can be reliably recorded. There are several other factors and considerations that will only reduce that capability.

In particular the system performance (MTF) of all interacting components is what is of real importance if we want to predict practical contrast/resolution performance. And even then there are other factors (e.g. flare resistance and bokeh) that determine quality, not to forget shooting technique (such as tripod/shutter-speed/aperture usage).

Bart

Marc Schultz said:

It used to be that people would talk about a minimum pixel pitch of 9 microns in order for a sensor to produce low noise, high quality pixels.

But then the envelope changed

Click to expand...

Even when the quality of pixels hits a wall, the quality of the image does not. The idea that the quality of an image is determined by the quality of individual pixels is the "flat earth" of current times; something that people repeat as if it were simply so, supported by conservative scientists, who miss the forest (image) for the trees (pixels).

Current compact cameras with 2 micron pixels image better per square mm than many current DSLRs do.

John_Nevill said:

Nicolas, will the event of a new 14mm, put the Sigma 12-24 in the shade? I guess the Sigma will still have the edge on low distortion (and of course price)

Click to expand...

Hi John
At least it shows that Canon has listended to the numerous complaints about ultra wide angles…

I still love my 12-24 but if the new Canon 14mm shows better sharpness in 4 corners and if distortion is at least eaqual to to the Sigma, I'll sign for it! don't forget it's f/2.8!

John Sheehy said:

Current compact cameras with 2 micron pixels image better per square mm than many current DSLRs do.

Click to expand...

I was thinking in terms of "better in every way". More specifically, only Canon DSLRs, at high ISOs, have less *image* noise than the best 2 micron compacts. All compacts are better for sensor MTF.

I always like these discussions, since I used to be an imagery analyst for the USAf at one time and did advanced sensor requirements for about 6 years.

I do know that there are a whole lot of factors that affect the final image quality, lens and sensor being only two of many general categories and those two have many factors affecting their contribution to the system as a whole.

I was told a few years back that 8um was 'bout' the optimum size for sensors below which tradeoffs started to occur as a serious complication, but I was also told that given the correct algorithms they expected that sensors might eventually drop to below 2.8um and still produce decent images at lower ISO levels, but not much farther then that before physics limited the noise in the images. What I was told seems to still hold true today.

Meanwhile I have put myself on the list for a 1DsMkIII because of multiple factors, including the fact I was told a 35mm sensor should be good to about 22-25MP before it would not be able to operate at most ISO settings. 21.1MP is a little below that, but I tend to think Canon is not going to be able to get beyond 25MP so if I get a 21.1MP sensor, I am probably golden for life in terms of 35mm DSLR.

The other factors are I have a 1DsMkII and would like a backup in the same class. I like the 14-bit sensor, although I would have preferred a 16-bit sensor - if one does come out though I am not trully expecting very many people to actually see the difference in terms of final output so can live with a 14-bit sensor.

The reason I am not getting a MF back (despite having a Hassleblad body and lens available are because all of the MF cameras I have researched are Studio or Landscape/still life cameras with limitations in terms of operating conditions for those things I have used my camera for over the years. No MF back that I am aware of can shoot in conditions where I need f2.0, ISO 1600 at 1/125th or worse light, because I can't use flash and I have been in conditions where it was marginal at f1.2, 1/80th at ISO 1600 (I love the newest ACR for fixing marginal shots )

trivia: I am looking forward to seeing if the quiet mode works better as it is pretty much useless on my 1DsMkII and almost couldn't shoot because the shutter recocking is highly distracting. One of the few reasons I have kept my 10D is because it is nearly silent when shooting.

Oh and try shooting 4FPS with a MF back (have been rare occasions for me)

So 1DsMkIII for me and I suspect that I will only see improvement in what i get out of my lenses based on what I know,

Thanks for putting up with my rambling,