No EOS 5DS & EOS 5DS R Replacements in 2017 [CR2]

[ahsanford]

b) you don't comprehend that ANY lens will show a benefit. lens performance such as with diffraction depends on your sensor to image magnification, not your pixel density.

I fully understand that even mediocre lenses will show a benefit from a higher resolution sensor, yes. But some 22 to 50 MP comparisons show that the degree to which a lens improves on the 50 MP canvas can vary quite a bit. On the 'lemon' lens tested at LR -- a beat up 50 f/1.4 USM -- it improved on the 50 MP sensor but not nearly as dramatically as the other higher quality lenses they tested.

And DXO, PZ would back that up. Some lenses that look quite good at 22 MP tend to show their age once they step up to 50 MP.

And to tr573's comment, I hear you -- I'm not saying any lens that 'only' scores 40 / 50 at DXO will be a train wreck at a higher resolution. Hardly! But some lenses that are up for 22 MP work but will not look much better than that on a higher resolution sensor, say an old 17-40L as an example (which did not make the cut for the 5DS recommended list). The same thing will be true in comparing lenses considered good enough for 50 MP work that won't do much better on a 120 MP sensor.

- A

 

[Maiaibing]

I don't see the point of using 120 Mp on a 35mm DSLR camera. (...) How many of you have a use for such a high res camera?

When did thousends of uers decide they needed a 3 MPIX camera, 5 MPIX camera, 8 MPIX camera, 10 MPIX camera, 12 MPIX camera, 15 MPIX camera, 18 MPIX camera, 20 MPIX camera .... 50 MPIX camera?

When they could see they could do more, better with more MPIX.

But I understand you are still shooting 3 MPIX?

I'll probably stand in line for having a chance to jump to 120 MPIX goodness.

 

[rrcphoto]

b) you don't comprehend that ANY lens will show a benefit. lens performance such as with diffraction depends on your sensor to image magnification, not your pixel density.

I fully understand that even mediocre lenses will show a benefit from a higher resolution sensor, yes. But some 22 to 50 MP comparisons show that the degree to which a lens improves on the 50 MP canvas can vary quite a bit. On the 'lemon' lens tested at LR -- a beat up 50 f/1.4 USM -- it improved on the 50 MP sensor but not nearly as dramatically as the other higher quality lenses they tested.

And DXO, PZ would back that up. Some lenses that look quite good at 22 MP tend to show their age once they step up to 50 MP.

again, no, you totally miss the point.

resolution is never the problem, the resultant image size is a problem. if you display a 50mp image on a 30" print and a 100MP image on a 30" print, it doesn't matter, the 100MP image will look better. each lens has an effective magnification ratio, where if you increase the image magnification beyond that, you will see signs of stress. it has nothing to do with pixel density. assuming a flat brick wall image, it entirely depends on where the loss of resolution is greater than the perceived circle of confusion and is visibly apparent as you move from areas of the image to other areas.

viewing at 100% magnification is not objective, nor does it even make sense and is most certainly not even real world.

 

[IglooEater]

I don't see the point of using 120 Mp on a 35mm DSLR camera. (...) How many of you have a use for such a high res camera?

When did thousends of uers decide they needed a 3 MPIX camera, 5 MPIX camera, 8 MPIX camera, 10 MPIX camera, 12 MPIX camera, 15 MPIX camera, 18 MPIX camera, 20 MPIX camera .... 50 MPIX camera?

When they could see they could do more, better with more MPIX.

But I understand you are still shooting 3 MPIX?

I'll probably stand in line for having a chance to jump to 120 MPIX goodness.

I agree that more Mpix are better, in there own right than less. While you have to admit that any average user can benefit from an upgrade from 3 MPIX, the upgrade from 50 to 120 is less compelling. Firstly, because it is a smaller upgrade, (only 2.1x more pixels instead of almost 7X more), and also because an actual use of the extra pixels demands a more specialized application such as heavy cropping, extreme print sizes, etc.

 

[rrcphoto]

I don't see the point of using 120 Mp on a 35mm DSLR camera. (...) How many of you have a use for such a high res camera?

When did thousends of uers decide they needed a 3 MPIX camera, 5 MPIX camera, 8 MPIX camera, 10 MPIX camera, 12 MPIX camera, 15 MPIX camera, 18 MPIX camera, 20 MPIX camera .... 50 MPIX camera?

When they could see they could do more, better with more MPIX.

But I understand you are still shooting 3 MPIX?

I'll probably stand in line for having a chance to jump to 120 MPIX goodness.

I agree that more Mpix are better, in there own right than less. While you have to admit that any average user can benefit from an upgrade from 3 MPIX, the upgrade from 50 to 120 is less compelling. Firstly, because it is a smaller upgrade, (only 2.1x more pixels instead of almost 7X more), and also because an actual use of the extra pixels demands a more specialized application such as heavy cropping, extreme print sizes, etc.

every print size will benefit from more Mp's unless you believe that oversampling does not lead to more accurate results.

I recall when canon back in the middle of the 2000's suggested that 60+MP cameras are achievable and canon would be doing them, and the internet world cried that there's no need for such MP's .. and that no one would be able to see the difference of said MP's...

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

 

[Antono Refa]

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

A 20x30" print would probably be viewed from around ~3ft (the diagonal of the print) or farther. For that viewing distance, you could print it at 180 DPI. So you'd need 20 * 30 * 180^2 = 19.5 MPIX, possibly less.

 

[Diko]

b) canon already stated 60 lenses are compatible with the upcoming 120MP camera. any lens will work on a 120mp camera that works on a lower resolving sensor.

By "any" - you are right. They said that. But how satisfactory the results would be.

Currently I don't want to shoot with 24-70L 2.8, since I can see with my eyes the difference on a regular 30MPish 5D4. The first version of that awesome lense delivers a visible blur. And that is after focus-calibration.

On the other hand the 70-200 2.8 L2 and Sigma 50 ART are completely other story. Recently haven't tested the 5Ds with the above mentioned, but pretty pretty sure I will not like the 24-70 any more. Waiting for some real reviews on the new Sigma 24-70 2.8 and back to back with the 24-70 2.8 Mark 2 from Canon.

...But something like 250MP would be horridness and not usable due to camera shake and diffraction cause by the atmosphere its self. This is why even Med Format does not go above 100MP at the moment. It becomes a point where more is not helping and its better for focus on quality other then quantity. ..

R sure about that? I mean MF not going above 100MPs?

I support the idea that there's no glass out there to support that many MPs on FF.

 

[symmar22]

I'm only passingly well read on diffraction, but I had a question.

I'm just curious as sensor resolution climbs and the diffraction limited aperture more and more approaches a wide-open aperture, how would landscapes work? It would seem very difficult to obtain peak sharpness given the larger DOF needs of most landscape compositions.

This is true of landscape work in general -- even on my 5D3 my lens might be sharpest at f/5.6 but I don't hesitate to stop down to f/11 or f/14 as needed. But on a massive 100+ MP canvas, I imagine you'd be throwing out a lot more detail to make a similar decision to stop down. So what is one to do in that case? Are people going to need to focus-stack their landscapes like a product/macro photographer would?

Please educate me here, this is not my wheelhouse at all. Thx.

- A

I work as an architecture and interiors photographer, and my experience with my 5DSr has removed all fears I had before. I use f11 to f18 - f20 on most of my pictures, and the sharpness just blows out my 5D2 at equivalent apertures. Diffraction is overrated IMO, when you need the f-stop, you just use it. A slightly OOF part of the image will be much more disturbing than a minor softness all over the picture. I did not replace any lens out of my collection, the sensor just seems to add resolution to all of them. You should avoid to stop down past f22, but that was already the case with my Nikon F4 in the film era.

There is a difference between pixel peeping for a hobby and actual work.

On top of that, for those who still wonder, I get much less moiré than I did with my 5D2, and much less noise as well since most of the final images are a bit down sampled anyway (few clients need actual 50Mp files), but the files are much cleaner on the 5DSr to begin with.

I did not change my way of working (tripod and live view), so the comparison is fair. The technical quality of my production has clearly improved with the new camera.

As a side note, I wonder how I could work with a blurry AA filter before, I hope the next iteration of the 5DS will be without any filter at all (for my type of work, one of the drawback of Canon cameras is the AA filters that are quite strong, and in my opinion do more damage to sharpness than stopping down to f11, and when you have strong moiré like fabrics, they are useless anyway).

Yes I would love a more modern equivalent sensor with more DR (on chip ADC) and less noise, but there is no hurry.

Finally, the only drawback is the slower editing time in LR and PS, but the bigger image allows more precise Photoshop retouching, when you deal at pixel level.

 

[Cthulhu]

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

A 20x30" print would probably be viewed from around ~3ft (the diagonal of the print) or farther. For that viewing distance, you could print it at 180 DPI. So you'd need 20 * 30 * 180^2 = 19.5 MPIX, possibly less.

This discussion makes me wonder if folks claiming you can't see the need for more mp have actually shot with a 5ds(r). It breathes new life into ANY lens I put in front of it. I tried one for a week and was never able to pick up my 5d3 again.
The image size alone is awe inspiring, excellent for cropping and prints. Sure you can print with less MP, but you will definitely be able to tell the difference.

As for camera shake, shooting a 5dsr is not any harder than shooting a 7dmk2 - which is to say not hard at all.

 

[ahsanford]

As for camera shake, shooting a 5dsr is not any harder than shooting a 7dmk2 - which is to say not hard at all.

Curious, has anyone ever done a hit-rate study on camera shake, best practice minimum shutter speed rules, etc. with 22MP vs. 50 MP?

I keep hearing motion will more harshly punish you on a higher resolution sensor, but has anyone ever shown how much this really matters? Do you need a different 1/FL rule for higher resolution sensors?

Have no idea on this, please let me know your thoughts.

- A

 

[tr573]

As for camera shake, shooting a 5dsr is not any harder than shooting a 7dmk2 - which is to say not hard at all.

Curious, has anyone ever done a hit-rate study on camera shake, best practice minimum shutter speed rules, etc. with 22MP vs. 50 MP?

I keep hearing motion will more harshly punish you on a higher resolution sensor, but has anyone ever shown how much this really matters? Do you need a different 1/FL rule for higher resolution sensors?

Have no idea on this, please let me know your thoughts.

- A

It's more that the more you magnify the image , the more motion will punish you. So if you're viewing a 22mp vs 50mp image both taken on 135 format sensors, both printed at 13x19, no difference. Same magnification.

But look at them on screen @ 100%, the 50mp image is magnified more, and you see the difference.

20MP APS-C vs 20MP 135 , you'll see the difference even in the print example, because the print from the APS-C sensor is magnified more.

 

[Antono Refa]

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

A 20x30" print would probably be viewed from around ~3ft (the diagonal of the print) or farther. For that viewing distance, you could print it at 180 DPI. So you'd need 20 * 30 * 180^2 = 19.5 MPIX, possibly less.

This discussion makes me wonder if folks claiming you can't see the need for more mp have actually shot with a 5ds(r).

There's a difference between *need* and *benefits from* higher resolution.

It breathes new life into ANY lens I put in front of it.

I'll bet that any includes only the sharper lenses with EF mount.

The image size alone is awe inspiring, excellent for cropping and prints. Sure you can print with less MP, but you will definitely be able to tell the difference.

Question is whether the difference I'll get with the lenses I have is worth the money.

I'm sure I'll get more details with the 70-200mm f/2.8L IS II USM, I doubt I'll get anything with the 24-105mm f/4L IS USM.

Personally, I'd rather invest my money in upgrading my EF 16-35mm f/2.8L mk2 to mk3.

 

[rrcphoto]

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

A 20x30" print would probably be viewed from around ~3ft (the diagonal of the print) or farther. For that viewing distance, you could print it at 180 DPI. So you'd need 20 * 30 * 180^2 = 19.5 MPIX, possibly less.

around yes, you do want to get around 200-300 dpi though for images, meaning that the 5Ds currently overs around 1:1 to 1:2 sampling for inspection. The 120MP version would oversample more. oversampling is a good thing.

however, your shutter speed (shake, motion,etc), lenses and your aperture is set by the print size, not by the amount of oversample.

 

[rrcphoto]

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

A 20x30" print would probably be viewed from around ~3ft (the diagonal of the print) or farther. For that viewing distance, you could print it at 180 DPI. So you'd need 20 * 30 * 180^2 = 19.5 MPIX, possibly less.

This discussion makes me wonder if folks claiming you can't see the need for more mp have actually shot with a 5ds(r). It breathes new life into ANY lens I put in front of it. I tried one for a week and was never able to pick up my 5d3 again.
The image size alone is awe inspiring, excellent for cropping and prints. Sure you can print with less MP, but you will definitely be able to tell the difference.

for sure, because the more Mp's the more you can oversample. those images will have better accuracy, tonality, less noise and better image acutance. Per pixel may look worse, but that's not the point.

 

[Cthulhu]

As for camera shake, shooting a 5dsr is not any harder than shooting a 7dmk2 - which is to say not hard at all.

Curious, has anyone ever done a hit-rate study on camera shake, best practice minimum shutter speed rules, etc. with 22MP vs. 50 MP?

I keep hearing motion will more harshly punish you on a higher resolution sensor, but has anyone ever shown how much this really matters? Do you need a different 1/FL rule for higher resolution sensors?

Have no idea on this, please let me know your thoughts.

- A

Pixel density is the matter. The 7dmk2 and 5dsr have practically the same pixel density.

Canon recommends 1/1.6fl for APSC and 1/2fl for the 5dsr. I shoot both the same as I shot my 5dmk3 and have had no issues.

 

[ahsanford]

Canon recommends 1/1.6fl for APSC and 1/2fl for the 5dsr. I shoot both the same as I shot my 5dmk3 and have had no issues.

Thank you -- had no idea Canon went on record about this.

For those who shoot auto ISO and set their shutter speed to auto (I recognize that's a small slice of us), I'm curious, do those 1/(1.6*FL) and 1/(2*FL) rules play out on the camera? Did Canon weave their guidance above into the camera's decision making processes?

- A

 

[Cthulhu]

Even if you only print 30x20" prints (which are probably the max for most people), each pixel being printed is oversampled the more MP's you have to work with. that's more accurate color and less noise, and less false sharpness.

A 20x30" print would probably be viewed from around ~3ft (the diagonal of the print) or farther. For that viewing distance, you could print it at 180 DPI. So you'd need 20 * 30 * 180^2 = 19.5 MPIX, possibly less.

This discussion makes me wonder if folks claiming you can't see the need for more mp have actually shot with a 5ds(r).

There's a difference between *need* and *benefits from* higher resolution.

It breathes new life into ANY lens I put in front of it.

I'll bet that any includes only the sharper lenses with EF mount.

The image size alone is awe inspiring, excellent for cropping and prints. Sure you can print with less MP, but you will definitely be able to tell the difference.

Question is whether the difference I'll get with the lenses I have is worth the money.

I'm sure I'll get more details with the 70-200mm f/2.8L IS II USM, I doubt I'll get anything with the 24-105mm f/4L IS USM.

Personally, I'd rather invest my money in upgrading my EF 16-35mm f/2.8L mk2 to mk3.

Anybody benefits from higher resolution - on a 5dsr you can print larger, with more detail, you can downsample and appear sharper or you can do extreme crops and still have perfectly detailed images, basically doubling your lens reach.

As for your bet, you'd lose:

https://www.lensrentals.com/blog/2015/06/canon-5ds-and-5ds-r-initial-resolution-tests/

 

[jeffa4444]

"Resolution, in line-pairs per distance, that can be achieved with a digital sensor is strongly constrained by the photosite frequency of the sensor. Even in the absence of blur from diffraction and lens aberrations, the highest line-pair frequency than can be obtained with high contrast will be only about one-fourth the photosite frequency. Therefore, resolution results reported in lens tests should be evaluated in light of the photosite frequency of the camera sensor. Maximum MTF50 resolutions that are substantially less than one-fourth the photosite frequency are suggestive of lens-limited systems. On the other hand, a maximum MTF50 resolution that is about one-fourth the photosite frequency is suggestive of a sensor-limited system. In order to refine the analysis, it is necessary to consider the effects of diffraction on contrast"

 

[jeffa4444]

"To add to the previous post we designed lenses for high end motion picture with strict photo site size tolerance" Thats not to say they cannot be used outside of this tolerance but they will not produce maximum resolution. Aberrations, motion blur and contrast will all affect apparent sharpness and lab test conditions are rarely achievable in the field.

 

[RGF]

I don't see the point of using 120 Mp on a 35mm DSLR camera. The slight vibration from the mirror slap and shutter movement would blur the image at the pixel level. Having to use mirror lockup on every shot kind of negates the reason to have a DSLR. Maybe a mirrorless studio camera would have some utility in the commercial market. I could also see some uses in robotic imaging for some manufacturing processes. How many of you have a use for such a high res camera?

At some point diffraction destroys resolution. I don't know the math so I can not say i will be f5.6 or something l like that. But between diffraction, slight camera vibration, less than stellar optics, ... not sure that images from a 120 MP camera would be noticeably better than 50 MP camera in all but a few limited scenarios.

Like to see Canon work on (make work) a foveon type sensor or reinvent the sensor totally to get better IQ (or effective resolution).

Not sure that we can get much more effective resolution from existing 35mm sensor technology with a major breakthrough or reinvention.

I'd love a Fovean style sensor in a Canon body but what are the issues behind the shutter speed dilemmas? Sigma bodies are infamous for not being able to shoot fast.

Do you mean FPS or minimum shutter speed such as 1/8000 of second?