High-resolution EOS R Camera, Where are you?

[Brikna]

I don't recall them making a big deal about it, though, and that sort of makes sense.

Maybe they didn't make a big deal out of it, but they certainly mentioned this new feature often and in many places.

Detailed Features

"...Reduced mirror vibration for maximum sharpness. The EOS 5DS R uses Canon’s Mirror Vibration Control System to reduce vibrations caused by internal movements within the camera, which can spoil image sharpness and reduce resolution. Cams are used to drive the mirror up and down in a more controlled fashion, avoiding any sudden stops and also softening the shutter-release sound. ..."

 

[roby17269]

No problem. Camera shake results from the sensor moving while the picture is taken. The amount of blur is a result of the distance the plane/angle of the sensor is moved during exposure. This is a simple geometrical function. The distance of movement to your sensor that induces motion blur does not change regardless if your camera has 1,000 or 100 mill pixels - the only factor is the distance the sensor moves while the picture is taken. Imaging holding a camera and tilting it 5 degrees down. No matter the size of your sensor, the light hitting that sensor will move precisely 5 degrees. So the blur from hand holding a low and high MPIX camera is always exactly the same. When I said it may be less with a high MPIX camera its because several high-MPIX DSLR-cameras have special dampened motors to reduce motion blur resulting from mirror slap.

The reasons why people think more pixels = more blur is probably 1) because when moving to high MPIX cameras people could suddenly "see" the motion blur, that they did not notice before, because sensor resolution was much higher - it was actually the same as before, just more visible when pixel peeping 2) both Nikon and Canon promoted their dampened motors as an advantage of high MPIX cameras - effectively establishing a link 3) dpReview (and a lot of others) when reviewing the first high MPIX cameras wrongly claimed that there was more motion blur - almost all serious reviewers have since retracted/corrected their initial mistakes. But just as the Force the Myth lives very strong on the net!

I understand what you say: the amount of blur is the same (assuming the amount of shake is the same, meaning same user, same fl, same IBIS/ILIS systems, same shutter mechanism). But the blur is more visible because of higher resolution assuming same pixel-level magnification.
In practical terms, what matters to me is the latter part. If I crop because I need / can (with a high-mp camera) then I am more likely to see more shake-induced blur. Which matches my experience where, all else being equal, I see more blur with higher resolution sensors.
Even if there is no causal relation, what matters to me is the practical fact that higher-mp cameras require more discipline shooting them or accepting more visible blur.

 

[neuroanatomist]

Maybe they didn't make a big deal out of it, but they certainly mentioned this new feature often and in many places.

Detailed Features

"...Reduced mirror vibration for maximum sharpness. The EOS 5DS R uses Canon’s Mirror Vibration Control System to reduce vibrations caused by internal movements within the camera, which can spoil image sharpness and reduce resolution. Cams are used to drive the mirror up and down in a more controlled fashion, avoiding any sudden stops and also softening the shutter-release sound. ..."

View attachment 212998View attachment 212999

Thanks!

Digging into it, the mirror vibration control system was introduced in 2014 with the 7DII, a 20 MP camera. It was used on the 5Ds (50 MP), 1D X II (20 MP), 5DIV (30 MP), 80D (24 MP), etc. So I'm still not seeing how it's associated with high MP sensors specifically, as @Fischer claims. Would be different if the feature was found only on the 5Ds/R, but rather it seems they introduced it on the 7DII and used it on xxD and higher bodies from then onward (until the end of new DSLRs).

 

[AlanF]

Thanks!

Digging into it, the mirror vibration control system was introduced in 2014 with the 7DII, a 20 MP camera. It was used on the 5Ds (50 MP), 1D X II (20 MP), 5DIV (30 MP), 80D (24 MP), etc. So I'm still not seeing how it's associated with high MP sensors specifically, as @Fischer claims. Would be different if the feature was found only on the 5Ds/R, but rather it seems they introduced it on the 7DII and used it on xxD and higher bodies from then onward (until the end of new DSLRs).

I used the dampened mode most of the time from the 7DII onwards as the mirror was noticeable at low shutter speeds. Now I use ES on the R7 to avoid shutter shock when necessary.

 

[vikingar]

Thanks!

Digging into it, the mirror vibration control system was introduced in 2014 with the 7DII, a 20 MP camera. It was used on the 5Ds (50 MP), 1D X II (20 MP), 5DIV (30 MP), 80D (24 MP), etc. So I'm still not seeing how it's associated with high MP sensors specifically, as @Fischer claims. Would be different if the feature was found only on the 5Ds/R, but rather it seems they introduced it on the 7DII and used it on xxD and higher bodies from then onward (until the end of new DSLRs).

EDIT: I think you can both be right.

Here's what Canon said for the 5DS and 5DSR:
https://www.canon.com.hk/cpx/en/technical/pa_EOS_5DS_and_5DS_R_Overview.html

50.6 Megapixels Ultra High Resolution image
EPS 5DS and 5DS R are built-in with a newly-developed ultra-high-pixel Full-frame Canon CMOS sensor with approximately 50.6 megapixel (...)

New designs and functions to ensure sharpness and details in ultra-high-pixel image
Camera-shake blur has a negative effect on image quality, and this effect becomes more apparent as the number of pixels increases. The main cause of camera-shake blur is the collision brought by the movement of the main mirror.

As a countermeasure, both 5DS and 5DS R adopt a new Mirror Vibration Control System (MVCS) which is a mirror control mechanism driven by motor and cams. This mechanism suppresses camera-shake blurring due to mirror motion in ultra-high-pixel image. Also, the rigidity of the tripod socket and surrounding parts have been improved to reduce camera vibrations caused by mirror motion.

Apparently the Fine Detail picture style was introduced for the same reasons:
https://www.canon.com.hk/cpx/en/tec...ess_parameters_for_Ultra_High_Resolution.html

New Picture Style “Fine Detail”
To maximize the resolving power and amount of detail captured by the 50.6-megapixel sensor, a new Picture Style “Fine Detail” has been developed and is first used in EOS 5DS and EOS 5DS R. In comparison to “Standard” and “Landscape”, “Fine Detail” emphasizes the gradation of tones and detail fineness of an image.

 

[neuroanatomist]

Here's what Canon said for the 5DS and 5DSR:
Camera-shake blur has a negative effect on image quality, and this effect becomes more apparent as the number of pixels increases. The main cause of camera-shake blur is the collision brought by the movement of the main mirror.

That was the first reason @Fischer gave, and IMO is really the only reason. Even then, it's only partially correct – it's not the number of pixels increasing, but rather the size of them decreasing that drives the increased manifestation of camera shake/vibration on an image (phrased differently, increasing MP count is only relevant in the context of the same size sensors).

The effects of camera shake/mirror slap on the 32 MP 90D are more evident than on the 50 MP 5Ds because the former has smaller pixels. The effects are about the same on the 7DII and 5Ds, because the pixels are almost the same size (but it was probably easier to develop the mirror vibration control for the smaller mirror of the 7DII).

 

[Walrus]

I used the dampened mode most of the time from the 7DII onwards as the mirror was noticeable at low shutter speeds. Now I use ES on the R7 to avoid shutter shock when necessary.

EFCS shouldn't already avoid shutter shock, without going to ES that cuts down to 12bit?

EDIT: given that R7 has EFCS, I guess it should have it...or, bringing along the M6II sensor, also brings along not having EFCS like the M6II?

 

[neuroanatomist]

EFCS shouldn't already avoid shutter shock, without going to ES that cuts down to 12bit?

I made a similar comment sometime back. It was then pointed out to me that the first curtain shutter shock on the R7 is sufficiently strong that it affects the next frame in a burst.

 

[vikingar]

That was the first reason @Fischer gave, and IMO is really the only reason. Even then, it's only partially correct – it's not the number of pixels increasing, but rather the size of them decreasing that drives the increased manifestation of camera shake/vibration on an image (phrased differently, increasing MP count is only relevant in the context of the same size sensors).

The effects of camera shake/mirror slap on the 32 MP 90D are more evident than on the 50 MP 5Ds because the former has smaller pixels. The effects are about the same on the 7DII and 5Ds, because the pixels are almost the same size (but it was probably easier to develop the mirror vibration control for the smaller mirror of the 7DII).

Agreed. The reasoning in the articles is mostly marketing, with some engineering tidbits made digestable for us common consumers. It's not like we get an interview with the actual engineers. The statements are not necessarily false, but they're sales pitches not academic papers.

I would guess engineering on a better mechanism started once higher-resolution / smaller-pixel bodies were on the horizon. Once developed, it probably made sense to use the same tech on all higher-end bodies regardless of resolution. Even just the smoother and quieter operation would justify it.

 

[Dragon]

For very sharp edges, I tend to rely on primes, never on ultra-wide zooms.
Yet, incredible progress has been achieved in designing WA zooms (EF 17-40 vs. EF 16-35 L III).

The problem with WA primes is that the selection of primes is very limited and also quite old for the most part, but the zooms are current designs and very good. I have found the RF 14-35 to be exceptional.

 

[Del Paso]

The problem with WA primes is that the selection of primes is very limited and also quite old for the most part, but the zooms are current designs and very good. I have found the RF 14-35 to be exceptional.

Leica M or Zeiss ZM WA primes also exist...
But at a cost.
And Canon's announced 14, 24 and 28mm primes wil, no doubt, be great lenses too!

 

[Dragon]

Logical, since that was Canon's own high MP DSLR. I don't recall them making a big deal about it, though, and that sort of makes sense. Never a good idea to highlight a previous problem with your technology unless you're planning to fix it across the board. They can tout the readout speed of the stacked sensor in the R3, but not by saying that's solution for the 'problem' of the slow readout speed of other MILCs.

Actually, they did make a pretty big deal out of it with drawings to show the controlled rates of acceleration and deceleration of the mirror. Rudy Winston did a whole video on the feature. The camera even has two settings for mirror acceleration rate and the slow one is very gentle (at the expense of a slower burst rate). Now that burst rates are through the roof, the shutter is as much of a problem as the mirror used to be. The M6 II was the first with a haybaler shutter, but not the last. That camera also has no EFCS option and the e-shutter is so crippled that it is not useful. I still like the camera, but I do wish Canon had offered the same EFCS option as the 90D (which also has a violent shutter).

 

[Dragon]

You agree then that when the 5DSR and R5 are tested for sharpness that is independent of the lens then the R5 just wins out. I did with the 5DSR have Moire occasionally spoiling details on birds feathers, especially Kingfishers tails for example, but not once has Moire interfered with the R5 output for me.

I don't think that is quite what I said. My experience is that the 5DSR actually has a better MTF than the R5 with the same lens, but the lack of AA filter can sometimes obfuscate that advantage due to aliasing if there is sufficient energy in the subject beyond the Nyquist limit. Another way to say that is if you were to test with a sinusoidal test pattern that stopped at the Nyquist limit, the 5DSR would have a higher MTF. In contrast, the Imatest approach looks at what is effectively a square wave input, so if the lens has adequate MTF, the sensor will definitely alias in the absence of an AA filter. Bottom line is that the difference is small enough that it is hardly worth arguing over, and if you have better results for your subjects with the R5 then it will be the obvious choice since it has so many other benefits in any case. It is interesting that Canon seems to be the only vendor left that is incorporating AA filters and that includes the MF vendors where aliasing can actually be a significant problem (although less so at 150 MP than at 50 MP or even 100 MP). It would be interesting to know if the decision to omit the AA filter by the other vendors was based on perceived image quality or just an attempt to save a buck.

 

[Dragon]

Leica M or Zeiss ZM WA primes also exist...
But at a cost.
And Canon's announced 14, 24 and 28mm primes wil, no doubt, be great lenses too!

Leica lenses don't fit my R5 and Zeiss lenses are MF . The RF 28mm pancake is excellent, but I don't see 28mm as particularly wide. The new lenses will likely be great (and expensive), but at this point they simply aren't there.

 

[scyrene]

Let me know when Canon releases that 2000mm or 2500mm focal length lens

RF 1200mm + 2x extender

 

[MJ]

The pixel pitch of my R7 is equivalent to an 84MP FF sensor. My EF Sigma 150-600 Sport lens works extremely well with the R7. The lens is something of a beast but that's mostly due to age.

Sorry, but isn't that 51.2 MP?

32MP x 1.6....

 

[Dragon]

Sorry, but isn't that 51.2 MP?

32MP x 1.6....

Nope, its an area function 1.6x1.6x32.3MP=82.7MP and there are some differing opinions re the MP count of the sensor since it is often described as a 33MP sensor, but 32.3MP is the size of the images.

 

[YuengLinger]

And I struggle to understand why so much MP is needed if is for the sake of cropping. That just means they are using wrong focal length.

I struggle to understand why anybody needs a zoom lens for the sake of framing. A good photographer would just choose the correct prime. Or zoom with their feet!

If it sells it sells.

 

[AlanF]

EFCS shouldn't already avoid shutter shock, without going to ES that cuts down to 12bit?

EDIT: given that R7 has EFCS, I guess it should have it...or, bringing along the M6II sensor, also brings along not having EFCS like the M6II?

You have missed the discussion on this we had here and is also on other sites. EFCS doesn’t affect the shot it is in but in a burst at lower fps, the mechanical closing of the shutter of one shot causes a vibration that affects the next shot on the R7.

 

[AlanF]

It is interesting that Canon seems to be the only vendor left that is incorporating AA filters and that includes the MF vendors where aliasing can actually be a significant problem (although less so at 150 MP than at 50 MP or even 100 MP). It would be interesting to know if that decision was based on perceived image quality or just an attempt to save a buck.

How does it save money by putting in an AA-filter in a newly designed sensor? (Not the 5DSR and 5DS compromise but a completely new sensor in a new camera). Filters presumably cost money so is that expenditure negated by a saving elsewhere?