EOS 5Ds- why all this negativity?

[Mark_Seng]

I really wonder if  the current canon l lenses (especially 17ts-e & 24 ts-e) resolve 50mp at common apertures like f8 - f13 ?
does someone have info on that?

[ErikKaffehr]

Hi,

Almost any good lens will resolve well on 50 MP, at least over the center of the field. The 7D has the same pixel pitch as the 5Ds, so resolution is like the 7D but the image is larger.

Lenses that are weak on the 5DIII will be as weak on the 5Ds, but if you pixel peek the difference will be more obvious. But, if you resize the images to a common size the 5Ds will always win.

Many wide angles have weak corners, and that will show even more with the 5Ds.

One more point, if you print same size, the 5Ds will always give a sharper image than the 5DIII, much or little sharper - that depends on the lens.

The 17 and 24 mm TS lenses are excellent lenses.

Best regards
Erik

[torger]

I really wonder if  the current canon l lenses (especially 17ts-e & 24 ts-e) resolve 50mp at common apertures like f8 - f13 ?
does someone have info on that?

Before I got a MFD tech camera back in 2012 I made a test with a Canon 7D and the ts-e 24 II, shifting the 7D lots of mm to simulate the edge of a full-frame sensor, the pixel density on the 7D corresponds to 45 megapixels. Unfortuantely the pixel crops are gone but my conclusion was that there was a much more significant sharpness falloff to the corner than you get on a MFD technical camera with say a Schneider Digitar 35mm (same field of view if you use a 49x37mm sensor).

So no you won't get MFD tech cam performance on the wide angles but the images will contain pretty high resolution anyway. In a way it's good when the sensor outresolves the lens, as then you get the maximum out of your lens investment.

Maybe the Zeiss Otus range will get a 24mm, and then I guess you would get MFD results in terms of corner to corner sharpness, but I don't dare to dream of the size weight and cost of that...

[ErikKaffehr]

Hi,

I don't argue with "Torger", just to mention that I have seen some images shot with the 17T&S on medium format cameras, 80 MP I believe. It was compared to the Rodenstock 23 HR. At f/8 there was no match, the "Rodie" won hands down, at f/16 it was quite even.

The way I see it, the 50 MP DSLRs, combined with the best available lenses, will be able to go into MFD territory.

To mention some examples:

- I have seen a comparison of Pentax 645 with their 24 mm lens compared to Zeiss 21/2.8 on the Sony Alpha A7r at Diglloyd, the A7r comfortably won.

- I have also seen a comparison of the Leica S2 and the Nikon D800 using the Zeiss 100/2.0 Macro Planar compared to the 120/2.5 Apo Macro Summarit. In that case the Nikon/Zeiss combo outperformed the Leica S2 MFD camera comfortably in the corners. In the central areas the Leica S2 was a bit better, but had excessive moiré. http://diglloyd.com/prem/prot/DAP/NikonD800/compare-LeicaS2-mosaic.html

Personally, I have none of these cameras. I am shooting a Sony Alpha 99 that has 6 micron pixels and a Sony Alpha 77 that has 3.8 micron pixels, pretty close to the Canon 5Ds, but at APS-C. I also shoot a lot with a P45+ back on a Hasselblad V-series camera and have some of Zeiss better lenses for that camera.

Getting back to the 3.8 micron Alpha 77 SLT, which has similar resolution to the Canon 5Ds at the pixel level. What I see is that some lenses perform very well. For instance, the 70-400 zoom is working very well at the short side. At 400 mm it is not really good enough. The 16-80/3.5-4.5 is quite decent at any focal length, even if the corners are lacking at short focal lengths.

In short, my take is this:

- You will get results that are as good as or probably better than 5DIII images with any lens.
- Really good lenses will be excellent overall

To mention a couple of examples:

- The old Canon 16-35/2.8 lens was a dog regarding corner sharpness. The new 16-35/4 is very good across the image. Just don't forget how many award winning shots have been made with that 16-35/2.8 lens…

- The 24-70/2.8L has been redesigned, the new lens is much superior to the old.

Best regards
Erik

Before I got a MFD tech camera back in 2012 I made a test with a Canon 7D and the ts-e 24 II, shifting the 7D lots of mm to simulate the edge of a full-frame sensor, the pixel density on the 7D corresponds to 45 megapixels. Unfortuantely the pixel crops are gone but my conclusion was that there was a much more significant sharpness falloff to the corner than you get on a MFD technical camera with say a Schneider Digitar 35mm (same field of view if you use a 49x37mm sensor).

So no you won't get MFD tech cam performance on the wide angles but the images will contain pretty high resolution anyway. In a way it's good when the sensor outresolves the lens, as then you get the maximum out of your lens investment.

Maybe the Zeiss Otus range will get a 24mm, and then I guess you would get MFD results in terms of corner to corner sharpness, but I don't dare to dream of the size weight and cost of that...

[Bart_van_der_Wolf]

- The old Canon 16-35/2.8 lens was a dog regarding corner sharpness. The new 16-35/4 is very good across the image. Just don't forget how many award winning shots have been made with that 16-35/2.8 lens…

Hi Erik,

Just for completeness, the Digital Photo Professional Raw (DPP) converter that comes with the Canon cameras, has an option called Digital Lens Optimizer (DLO). It allows to considerably improve the raw conversions with DPP from files that were show with that lens. One has to download the correction data file for the specific lens that one would like to add to the list of options.

I've tried it on my shots with the 16-35mm f/2.8, and the corners were significantly improved, they became quite usable actually.

By now, most Canon lenses have correction data available, including for the combined use of lenses with focal length extenders. Unfortunately it only works with DPP conversions, since the corrected Raw data is added to the original Raw data (which doubles the file size), in order to keep the Raw file also compatible with other converters (which are oblivious to the existence of two sets of Raw data).

Cheers,
Bart

[BJL]

I really wonder if  the current canon l lenses (especially 17ts-e & 24 ts-e) resolve 50mp at common apertures like f8 - f13 ?

Here is an experiment that someone with the right lenses could try.  Put those Canon lenses on a Canon _film_ camera, load it with a very high resolution low speed monochrome film like Ilford Delta 100 or Kodak TMAX-100, photograph your favorite test chart or brick wall, and examine the results.  Some of these monochrome films have higher resolution by any measure than even these 50MP sensors. And I mean useful resolution, with a healthy MTF like 50% or better, not just the extinction resolution (MTF about 7%) or measurements with high contrast 1000:1 test charts, as quoted by some "film still rules" partisans.

[dwswager]

It handily beats the old Canon 24-70 f/2.8.

But the 24-70 f/2.8 II is prime-sharp corner-to-corner and has minimal CA.

Nikon is yet to release a version competitive with this.

The current Nikkor is softer in the corners wide open, but that corner softness disappears by f/4.  And it is is slightly sharper in the center than the NEW Canon 24-70mm f/2.8 USM L II.  And the Nikon does exhibit approximately double the CA of the minimal CA of the Canon.   Bottom Line: In the real world, these lenses are basically equivalent in performance.  Nikon has not released a newer version because this lens set the standard when it was released and it is still a great performer.  Of course, Rumors of a PF/VR versions abound.

[alan_b]

I really wonder if  the current canon l lenses (especially 17ts-e & 24 ts-e) resolve 50mp at common apertures like f8 - f13 ?
does someone have info on that?

TSE 17 & 24 already show softness when shifted on 36MP, whether it's a problem is up to the user/application. 

[uaiomex]

I certainly fall into that camp! Sony's solutions are looking more interesting by the day.

+One

[bokehcambodia]

  Canon PR admitting to segmentation tactics...

"A Canon representative told DP Review that the ISO cap is completely "arbitrary".

In a similar vein, Canon has probably restricted the two new cameras to 5 fps continuous (the Mark III can do 6 fps)—and for some reason, the new cameras can't do uncompressed HDMI output, while the Mark III can. These differences might seem fairly insignificant, but if you look at the entire Canon DSLR lineup, there's an awful lot of feature fragmentation. That's great from a business perspective (product segmentation increases sales), but not so good for consumers, who end up having to buy more than one camera to fulfill all their needs."

http://arstechnica.com/gadgets/2015/02/canon-reclaims-resolution-crown-with-50-megapixel-5ds-and-5ds-r/

[bokehcambodia]

Good point, reading the past Canon rants (on the money) from Michael, i bet he has something to say...

Or LuLa may simply ignore the camera - which is entirely possible (the 5DIII was ignored.)

What would be worse for the 5Ds - LuLa ignoring it or LuLa panning it for "same IQ but more of it"?

[marc aurel]

I really wonder if  the current canon l lenses (especially 17ts-e & 24 ts-e) resolve 50mp at common apertures like f8 - f13 ?
does someone have info on that?

Hi Mark,
I have tested that for 36MP on the A7R (http://www.luminous-landscape.com/forum/index.php?topic=93273.0). A few megapixels less, but it should help to see, what you can expect. I think that most part of the image circle will resolve 50MP quite well. The extreme corners when fully shifted are a different story.
Best regards - Marc

[Bart_van_der_Wolf]

The current Nikkor is softer in the corners wide open, but that corner softness disappears by f/4.  And it is is slightly sharper in the center than the NEW Canon 24-70mm f/2.8 USM L II.  And the Nikon does exhibit approximately double the CA of the minimal CA of the Canon.   Bottom Line: In the real world, these lenses are basically equivalent in performance.

Hi,

Your charts do not specify, but I assume they depict Line Pairs per Picture Height (LP/PH) scores. The big question then becomes, on which camera body? If the Picture Height is different, then we are comparing apples with oranges ...
For a fair comparison between different sized sensors, one must compensate the LP/PH scores by the ratio of the sensor heights, so e.g. 4912/5792=84% for the smaller MP sensor scores.

Also, a 50.3 MP sensor will roughly pull up to 10% more (MTF) contrast out of the same lens at certain given absolute cycles/mm resolution than a 36.2 MP sensor of the same dimensions does. This will translate into better S/N and more usable data for post-processing, with more room to improve quality by down-sampling.

Cheers,
Bart

[dwswager]

Hi,

Your charts do not specify, but I assume they depict Line Pairs per Picture Height (LP/PH) scores. The big question then becomes, on which camera body? If the Picture Height is different, then we are comparing apples with oranges ...
For a fair comparison between different sized sensors, one must compensate the LP/PH scores by the ratio of the sensor heights, so e.g. 4912/5792=84% for the smaller MP sensor scores.

Also, a 50.3 MP sensor will roughly pull up to 10% more (MTF) contrast out of the same lens at certain given absolute cycles/mm resolution than a 36.2 MP sensor of the same dimensions does. This will translate into better S/N and more usable data for post-processing, with more room to improve quality by down-sampling.

My point was simply to refute the often posted comment that because the Nikkor has not been updated since it's introduction 8 or 9 years ago, the Canon is superior when, in fact, they perform so similarly that they are basically identical.  While Canon failed at their 1st attempt, Nikon didn't.  There has not been a need to update it and doing so only to add VR would have added cost, weight and size to and already expensive, big and heavy lens.  I've discounted the rumors about an update until the PF patent was discovered.  If Nikon releases a PF version update to this lens that maintains current optical performance (Big if) while substantially reducing the size and weight while adding VR it will sell like hotcakes!

The Nikkor was tested on the D3x, but Photozone did not specifically indicate on which body the Canon was tested.  However, they tend to test on like platforms.   And they warn that tests are not directly comparable between systems.  I've never seen a photoszone lens test with a 36MP Nikon on Photozone.  And I chose Photozone because Canon users seem to have a big chip on their shoulders concerning DxOMark.

[dwswager]

5Ds: 5 fps @ 50MP is 250MP/sec (2 x DIGIC 6)
5D3: 6 fps @ 22MP is 132MP/sec (1 x DIGIC 5+)
7D2: 10 fps @ 20MP is 200MP/sec (2 x DIGIC 6)
1DX: 12 fps @ 18MP is 216MP/sec (2 x DIGIC 5+)

I don't see the FPS as being a limitation that is arbitrary.

And it doesn't matter anyway.  The 5Ds is not intended as either a sports, wildlife or general use camera.  Though 5fps is still functional for those uses, just not optimal.  The target audience is fairly small.  Thinking 5fps and an ISO that topped out at 400 would probably been sufficient for the vast majority of the target market.  

That aside, it is frustrating when camera makers do arbitrarily cripple cameras.  Don't think it bothers most professionals or people looking at a camera for a specific purpose, but for me, as an amateur looking for a general use camera it hurts.  The small buffer on the D7100 turned what could have been a great general use camera into merely good for sports.  I can shoot the D810 in 1.2x crop mode, get the same 24MP image and 6fps, but overall performance is much better due to better data throughput and larger buffer.

[shadowblade]

My point was simply to refute the often posted comment that because the Nikkor has not been updated since it's introduction 8 or 9 years ago, the Canon is superior when, in fact, they perform so similarly that they are basically identical.  While Canon failed at their 1st attempt, Nikon didn't.  There has not been a need to update it and doing so only to add VR would have added cost, weight and size to and already expensive, big and heavy lens.  I've discounted the rumors about an update until the PF patent was discovered.  If Nikon releases a PF version update to this lens that maintains current optical performance (Big if) while substantially reducing the size and weight while adding VR it will sell like hotcakes!

The Nikkor was tested on the D3x, but Photozone did not specifically indicate on which body the Canon was tested.  However, they tend to test on like platforms.   And they warn that tests are not directly comparable between systems.  I've never seen a photoszone lens test with a 36MP Nikon on Photozone.  And I chose Photozone because Canon users seem to have a big chip on their shoulders concerning DxOMark.

Never realised the Nikon was that sharp.

I wonder if they'll release a new version to fix the CA and make it even sharper. Don't need VR for these focal lengths - would rather they made it as aberration-free as possible.

[Bart_van_der_Wolf]

The Nikkor was tested on the D3x, but Photozone did not specifically indicate on which body the Canon was tested.  However, they tend to test on like platforms.   And they warn that tests are not directly comparable between systems.

In that case we are probably comparing a Canon 5D Mark II (3744 px high) with a Nikon D3X (3744 px high), according to their Imatest chart legends, which should be reasonably well comparable except for strength of AA-filter, micro-lens design, and sensel aperture.

Modern lenses have the filter stack of digital cameras incorporated in the optical design. That's one of the reasons that modern lenses perform better than those from the film era. Also computer design has improved and the choice of lens materials is different and the use of aspherical lens elements has increased since they can be produced faster on a larger scale these days.

Cheers,
Bart

[ErikKaffehr]

Hi,

Lens rentals compared 24-70/2.8 alternatives from Canon, Nikon and Tamron.

What he found was that the Canon 24-70/2.8L II was the best of three bys some margin when tested on the MTF bench, but when the lenses were tested on the Nikon D800E and the Canon 5DIII the Canon lens was left behind. Why, because the better MTF of the Canon could not compensate for the superior resolution of the Nikon.

The article is worth reading: http://www.lensrentals.com/blog/2013/01/a-24-70mm-system-comparison

Best regards
Erik

In that case we are probably comparing a Canon 5D Mark II (3744 px high) with a Nikon D3X (3744 px high), according to their Imatest chart legends, which should be reasonably well comparable except for strength of AA-filter, micro-lens design, and sensel aperture.

Modern lenses have the filter stack of digital cameras incorporated in the optical design. That's one of the reasons that modern lenses perform better than those from the film era. Also computer design has improved and the choice of lens materials is different and the use of aspherical lens elements has increased since they can be produced faster on a larger scale these days.

Cheers,
Bart

[AlterEgo]

but Photozone did not specifically indicate on which body the Canon was tested.

they did - it is shown on the charts , top left corner = 5D mkII and also they always show the camera on the first page of the test report with the lens mounted

[dwswager]

they did - it is shown on the charts , top left corner = 5D mkII and also they always show the camera on the first page of the test report with the lens mounted

On the nikon it is specifically stated in the 1st or 2nd paragraph.