5DS and 5DS R Sensor Scores at DXO

[StudentOfLight]

Interesting! According to DXO, the Canon EF 300mm f/2.8L IS II USM resolves 20.5mp mounted to a Canon 6D!! The sensor is only 20.2mp, so I guess it manages to increase the camera's resolution somehow...seems legit.

You are quoting the wrong resolution. Actually, the 6D sensor is 5562x3708... that comes to 20.6MP but pixels are blanked off (I think for functions like peripheral falloff correction and noise reduction). The resulting images one gets from the 6D are only 5472x3648 which comes to 19.97MP. So The 300mm f/2.8 L II is even better than you thought.

 

[Neutral]

Do you really want to tell that 1dx or any other DSLR servo AF is able to track objects on the frame area where there is no focusing points ? If so then how it is done in the frame corner where here is no focusing points ?
Or you want to tell that 1DX has focusing points area covering the whole frame?
That all above (that it is possible to focus without focusing points) is really interesting and if true then this would be real revolution for DSLR focusing in normal mode ))) .
This would be dream for many people.
I am not talking about live view on 1DX which I find so much inconvenient and cubesome compared to a7s and a7r. Never use live view on 1DX. I am talking about normal DSLR shooting mode using mirror and separate AF sensor.
I have both Canon1DX and Sony a7 and can compare both and use each one for which it is suited better.
1DX and a7 complement each other very well for different situations.

It depends on what you mean by "no focusing points area." The actual AF sensor is much larger than the rectangle you see, so I have grabbed subjects well to the left or right of the outer points.

But if you mean in a true sense, true edge to true edge, then I can be on board with that. I would appreciate coverage out farther to the edge of the frame.

Definitely, I meant “true sense” - true edge to edge .
This was my dream for very long time to have AF sensor on my camera with AF points covering the whole frame – corner to corner. In addition, even more – MF camera (80-100) mpx with AF points covering the whole frame. This could make some sports photography disciplines much easier to shoot – easier to follow object.
I have no doubt that with 1DX you can get objects grabbed in the areas outside focusing points area but under some specific condition – if you already was locked on focus using one of focusing points and then system was tracking locked object . Or if the new object is entering the frame at the same distance with previously tracked object.
1DX is using intelligent AF tracking system to make AI Servo AF focus more accurate.
Intelligence means that AF does some object trajectory and speed predictions based on gathered tracking information so far. Therefore, it is possible to extrapolate object position and object speed for object leaving focusing points area.
What it is not possible to do with 1DX is to start focus on object from the corner or edge of the frame where there is no focusing points and continue to track it until next corner.
It is only possible to start AF tracking using one of the existing AF points and system in some conditions could predict (extrapolate) object position after it leaves focusing area with focusing points. More far extrapolation goes - more extrapolation error is accumulated.
Though A7S have much slower continuous AF than 1DX with dedicated AF sensor but it can grab object in AF lock box anywhere in the frame – even at corner, adjust AF box size to match the locked object size, and track this object across the frame up to the next corner. Like what you see on the jet fighter pilot weapon targeting system. He find the target, get it locked in the tracking box and system follows the target. Then when pilot find the best moment he presses shoot button.
With fast technology evolution, these technologies are now coming to the consumer market. Hope that soon we will see some significant progress on this in the FF ML cameras. It is only matter of sensor technology, sensor read speed, onboard processing power, and use of more sophisticated AF algorithms enabled by more onboard processing power in camera. Mirror and separate AF sensor will become obsolete when ML AF speed will match dedicated AF speed on DSLRs. ML is the future coming in reality now..

 

[Neutral]

You just opened up the A7 secret... ;D I do get missed focus shots with that.

On the other side, the Eye Focus is excellent. Shooting with FE55 @ f1.8, I can say my rate is 99.9%. Can't wait to hand on the new Batis 85mm and 25mm.

1Dx has Face focus feature in LiveView, but much slower than Eye Focus in A7.

There is no perfect camera. I would care less what others have to say about mirrorless. I enjoy open my camera closet and pick out proper tool for my shooting. In another words, why not enjoy the best from both worlds

I think we are both in the same boat and having basically the similar set of cameras/lenses, the same for some others here.
If it is possible to afford the best from both worlds,why not to do so ?
Live is short and passes quickly so it is better not to be too late to enjoy the best things there )))

My motivation to start with a7R was my Canon EF 17TSE which I like but it was extremely inconvenient to use it on 1DX. Also I was tired to wait for High Res body from Canon.
So I bought a7R for use as compact digital back for 17TSE and EF 24-70 F2.8 USM II and I was not disappointed.
Then some time later I could not resist buying A7S and more I used it more I liked it.
Especially after getting latest Zeiss FE 35 F.1.4 ZA.

 

[Loswr]

Nobody would suggest the 600 for indoor snaps of the family at home...

Well, maybe no one would suggest it, but that doesn't mean it won't work for that use case...

EOS 1D X, EF 600mm f/4L IS II, 1/160 s, f/4, ISO 10000

 

[mackguyver]

Nobody would suggest the 600 for indoor snaps of the family at home...

Well, maybe no one would suggest it, but that doesn't mean it won't work for that use case...

EOS 1D X, EF 600mm f/4L IS II, 1/160 s, f/4, ISO 10000

That's a mighty big room or long hallway...cool example.

 

[StudentOfLight]

Nobody would suggest the 600 for indoor snaps of the family at home...

Well, maybe no one would suggest it, but that doesn't mean it won't work for that use case...

EOS 1D X, EF 600mm f/4L IS II, 1/160 s, f/4, ISO 10000

Perhaps this is a case where a 50mm f/1.8 would be "better" than the 600mm f/4 ;D

 

[3kramd5]

Nobody would suggest the 600 for indoor snaps of the family at home...

Well, maybe no one would suggest it, but that doesn't mean it won't work for that use case...

EOS 1D X, EF 600mm f/4L IS II, 1/160 s, f/4, ISO 10000

HAH! Nice. Still, nobody would suggest BUYING one for that purpose. On the other hand, I have a camera shy son. Maybe I should build a deer blind in the living room and camp out with a supertele

 

[LonelyBoy]

Seems to back up the group thinking here where everyone desires faster lenses over slower lenses.

All else equal, are you telling me you'd prefer the slower lens?

 

[mangobutter]

Obviously a lot of DXOMark haters happen to be Canon fanboys. I'm a Canon shooter myself and even I recognize their sensors are not on par with others. They aren't as sharp and the dynamic range isn't all that great. This is easily shown in objective testing and not so easily in subjective testing. It's no secret that Canon has recycled their sensors for years and has responded with the new 24MP crop sensor and the new 50MP full frame sensor. Maybe their approach is conservative and concentrated on glass instead (which shows). Because Nikon lenses kind of suck.

In the past 5 years, Canon sensors "suck" but good enough to sell and create beautiful photos. Why use resources on developing new ones? Now we are to a point where more resolution is driving the market and there's an actual need for new sensors. Now we have the Rebel T6, 5DS/R, and 7DMII

 

[YellowJersey]

Obviously a lot of DXOMark haters happen to be Canon fanboys. I'm a Canon shooter myself and even I recognize their sensors are not on par with others. They aren't as sharp and the dynamic range isn't all that great. This is easily shown in objective testing and not so easily in subjective testing. It's no secret that Canon has recycled their sensors for years and has responded with the new 24MP crop sensor and the new 50MP full frame sensor. Maybe their approach is conservative and concentrated on glass instead (which shows). Because Nikon lenses kind of suck.

In the past 5 years, Canon sensors "suck" but good enough to sell and create beautiful photos. Why use resources on developing new ones? Now we are to a point where more resolution is driving the market and there's an actual need for new sensors. Now we have the Rebel T6, 5DS/R, and 7DMII

The 5DS/R are certainly interesting developments. The sensors seem to be a big improvement over what Canon has been putting out the last few years, too. I'm curious to see where all this goes. The 5DmkIV will probably be a big indicator as to whether Canon has really got back into the sensor game to the point where it can compete with Sony on the sensor side of things. I wouldn't say Canon has to beat Sony (though it would be nice), but so long as the difference isn't big, then it would be acceptable.

I think people also forget that there's more to a camera than the sensor. The sensor is a major part of it, don't get me wrong, and probably the most important part, but there are other things that can make or break a camera in the eyes of a potential buyer. I think Canon may know this too, as they seem to have been focusing on autofocus a lot (no pun intended.... ok, I lie, pun totally intended). This shouldn't be an excuse for poor sensor tech, but still worth keeping in mind. After all, the A7r has an amazing sensor, but there are some other aspects that are deal breakers to some people, such as poor autofocus, loud and clunky shutter, poor battery life, no optical viewfinder, lossy RAW files, and so on. Some of these have been resolved (see what I did there?) with the A72. This isn't to excuse Canon's shortcomings (they've done some REALLY annoying things), but more to explain that there's no "perfect" camera and a lot of it boils down to what you value most. I like some things about Canon, and dislike others. Same goes for Nikon, Sony, and Pentax (Pentax, imho, doesn't get nearly enough credit). [/soapbox]

One thing I've been wondering for a while now is if Canon is/was capable of making sensors that could go toe to toe with Sony. Maybe they focused on autofocus and other things for so long because they just didn't know how to make sensors like that? Magic Lantern probably discredits this notion a bit, but it does make me wonder if Canon wanted to make better sensors but couldn't figure out how.

 

[dcm]

The 5Ds twins look quite interesting, independent of any DxO sensor scores including Dynamic Range (DR).

I've waded through some of the DR discussions in the last year, mostly for amusement. There seem to be 3 camps that compare DR differently based on their perspectives and needs. The charts below should help illustrate these 3 groups. It's all about where you set the baseline and the topline.

The first group using Comparison 1 is strictly focused on the difference and closing the perceived gap. The baseline is the lowest current score and the topline is the current best score. Any difference is considered significant and the difference would help them in some current shooting circumstances. DxO often uses this perspective in their graphs to highlight the differences.

The second group using Comparison 2 takes a broader view and considers the complete, current picture. The baseline is zero (or 1) and the top line is the current best score. Here the perceived difference is much smaller when considering the total amount of DR provided. This is why many find Canon sufficient for their needs - they consider the gap to be an incremental improvement that doesn't afford them all that much. Other things bother them more.

The last group uses Comparison 3 to take a long term view of where DR might go in the future to meet our needs. The baseline is zero (or 1) and the topline is well above the current best. I chose 32 bits as the topline for a reasonable target to achieve with future technology. This group includes folks that find none of the current sensors sufficient for their DR needs and will shoot HDR to overcome deficiencies with the current best sensors. A significant improvement might be double the current DR for this group depending on their HDR bracket range. The difference in current sensor technology isn't worth mentioning to those of us in this group.

BTW: Edward Tufte offers some great insights on the design of graphs in Visual Display of Quantitative Information and his subsequent books.

 

[3kramd5]

snip

Count me in camp two, although showing it linearly instead of in stops may be the most genuine.

 

[dcm]

snip

Count me in camp two, although showing it linearly instead of in stops may be the most genuine.

DxO quotes and compares it in stops so I chose to use the same method for consistency.

 

[3kramd5]

snip

Count me in camp two, although showing it linearly instead of in stops may be the most genuine.

DxO quotes and compares it in stops so I chose to use the same method for consistency.

Right, I'm just saying the the unit of measure also skews the chart. I could express dynamic range in powers of 100 and show what is for all intents and purposes a flat line.

 

[StudentOfLight]

The 5Ds twins look quite interesting, independent of any DxO sensor scores including Dynamic Range (DR).

I've waded through some of the DR discussions in the last year, mostly for amusement. There seem to be 3 camps that compare DR differently based on their perspectives and needs. The charts below should help illustrate these 3 groups. It's all about where you set the baseline and the topline.

The first group using Comparison 1 is strictly focused on the difference and closing the perceived gap. The baseline is the lowest current score and the topline is the current best score. Any difference is considered significant and the difference would help them in some current shooting circumstances. DxO often uses this perspective in their graphs to highlight the differences.

The second group using Comparison 2 takes a broader view and considers the complete, current picture. The baseline is zero (or 1) and the top line is the current best score. Here the perceived difference is much smaller when considering the total amount of DR provided. This is why many find Canon sufficient for their needs - they consider the gap to be an incremental improvement that doesn't afford them all that much. Other things bother them more.

The last group uses Comparison 3 to take a long term view of where DR might go in the future to meet our needs. The baseline is zero (or 1) and the topline is well above the current best. I chose 32 bits as the topline for a reasonable target to achieve with future technology. This group includes folks that find none of the current sensors sufficient for their DR needs and will shoot HDR to overcome deficiencies with the current best sensors. A significant improvement might be double the current DR for this group depending on their HDR bracket range. The difference in current sensor technology isn't worth mentioning to those of us in this group.

BTW: Edward Tufte offers some great insights on the design of graphs in Visual Display of Quantitative Information and his subsequent books.

Your point is very well illustrated.

 

[caMARYnon]

The first group using Comparison 1 is strictly focused on the difference and closing the perceived gap. The baseline is the lowest current score and the topline is the current best score. Any difference is considered significant and the difference would help them in some current shooting circumstances. DxO often uses this perspective in their graphs to highlight the differences.

The second group using Comparison 2 takes a broader view and considers the complete, current picture. The baseline is zero (or 1) and the top line is the current best score. Here the perceived difference is much smaller when considering the total amount of DR provided. This is why many find Canon sufficient for their needs - they consider the gap to be an incremental improvement that doesn't afford them all that much. Other things bother them more.

The last group uses Comparison 3 to take a long term view of where DR might go in the future to meet our needs. The baseline is zero (or 1) and the topline is well above the current best. I chose 32 bits as the topline for a reasonable target to achieve with future technology. This group includes folks that find none of the current sensors sufficient for their DR needs and will shoot HDR to overcome deficiencies with the current best sensors. A significant improvement might be double the current DR for this group depending on their HDR bracket range. The difference in current sensor technology isn't worth mentioning to those of us in this group.

Very well said.

 

[jeffa4444]

Here is how we test lenses and then lenses on cameras:-

1. Each lens is measured on an MTF machine on axis and off axis at six pre-determined points, the measurements are averaged and then compared to theoretical numbers. The results are recorded by serial number, date & time.
2. Each lens f stops is measured on an f stop machine (this also reads T stops).
3. Each lens is projected and checked at 6ft & infinity and to see various chromatic abberations against a standard graticule. Its also checked for distortions. For zooms lenses they are checked at each focal lengh and for tracking. Experiance tells you about various flaws like color fringing, fall off, veiling glare, contrast etc.
4. Once we understand the characteristics of the lens we then shoot tests on a camera at different f stops again at 6ft & infinity. 6ft against a chart and infinity against a brick wall we use for all infinity tests. We do this more for practical testing than theoretical testing.
5.Sometimes we shoot harp tests to check for forward or back focus (particularly important when shooting video with splits)
6. Cameras are mounted on lens test posts NOT tripods

Cameras:-
1.We check the focal depth, speeds and dynamic range.
Dynamic range is tested with a special lens and shooting a sphere that is specially designed in Germany for measuring Dynamic range up to 16 stops.

Our goal is to check lenses, check cameras and then check the two together its a bit more involved than Ive explained here (lighting for example) but infinately more concise than DXO.

Many professionals are NOT looking for more resolution or more perfect images they often want certain imperfections to enhance the subject matter such as older uncoated lenses, lenses with flares of a certain type. I always say you can break a picture down but not make a bad picture good & its also why in B&W film has made a comeback.

 

[jeffa4444]

We also now know DXO are no longer neutral pitching the DXO One as a DSLR alternative.