Canon EOS-1D X DXOMark Sensor Scores

[thepancakeman]

i guess for some the reputation of the DXOmark has just become a bit better. :

;D ;D
I think this is the first time admin put Canon's DXO mark on the front page

Yeah, I thought he said earlier this year that he was never going to post anything dxomark-related on the front page ever again because their sensor scores were meaningless.

But see, this isn't from DXO, it's a rumor about DXO, so it's all good. ;D ;D

 

[sanj]

What's interesting is that those numbers suggest that the 1D X sensor trumps all other sensors in terms of low-light performance. In fact, based on their numbers they are saying there is a 2x difference in improvement in low-light ISO between the 1D X and the 5D3, than between the 5D3 and 5D2. I just don't see it.

Is that what they saying? I do not see it either.

 

[Chewy734]

Is that what they saying? I do not see it either.

Well, the low-light ISO scores are 3296, 2293, and 1815, for the 1D X, 5D3, and 5D2, respectively.

The difference between the 1D X and 5D3 is 1003, and the difference between the 5D3 and 5D2 is 478. So, the difference between the former is ~2x as much as the latter.

Or am I missing something here?

 

[tron]

I will believe these numbers when I see them on the DXO site itself.

Come to think about it: some folks who tried to lift the shadows in 1DX images do not find the performance any better than the 5D3... Hmmm...

Now this is very interesting information, especially for non-sports photographers...

 

[preppyak]

The difference between the 1D X and 5D3 is 1003, and the difference between the 5D3 and 5D2 is 478. So, the difference between the former is ~2x as much as the latter.

Or am I missing something here?

Well, you're thinking in terms of direct numbers, whereas ISO isn't proportioned that way. The difference between 1600 and 3200 is the same as between 3200 and 6400, despite the later having twice as great a difference in linear terms. So, really what it is saying is the 5DII and III are pretty close in ISO performance, but the 1DX is another step above.

This is how DxO mark explains it

Sports Score is based on Low-Light ISO performance (values in ISO index). Low-Light ISO indicates the highest ISO sensitivity to which your camera can be set while maintaining a high quality, low-noise image (based on a Signal-to-Noise-Ratio [SNR] of 30dB, a dynamic range of 9EVs and a color depth of 18bits). As cameras improve, the highest ISO setting to produce 30dB, 9EVs, 18-bit images will continuously increase, making this scale open. Low-Light ISO performance is of primary importance in photojournalism, sports and action photography.

So, they are still following the ISO index, which is based on doubling the previous number being equal to 1-stop. So, for example, the difference between the 5dII and 1DX, by their score, is essentially 1 stop. The difference between the 5dII and 5dIII is 1/3 stop. But, what they are actually measuring for is their own benchmark, which is kind of arbitrary. It's not saying the 1DX is a whole stop better than the other cameras; it's saying its a stop better by their definition of "good". If your standards aren't 30dB, 9EV, 18-bit images, then you might find completely different numbers for the difference between cameras.

That's why its important for an objective test to state what they are measuring, that way you can decide whether their measurement is important to you or not. Someone who cares less about dynamic range but more about lower read noise might get very different results. Likewise, someone shooting in B+W might care way more about DR and not so much about noise/grain

 

[Chewy734]

Thanks for the explanation preppyak.

 

[jrista]

Everyone needs to keep in mind, these are not yet posted as official results. Assuming they ARE official results, something else to keep in mind:

A DR of 12.8 is the Print DR. As everyone here who has read the debates about DXO before, particularly from Neuro and myself, Print DR statistics from DXO are very misleading. These numbers are not "impossible" like the D800's 14.4, however to remain objective and consistent in my argument:

Just as you couldn't actually capture a scene with 14.4 literal stops of DR with a D800 IN-CAMERA, neither will anyone be able to capture a scene with 12.8 literal stops of DR with a 1D X. For the exact same reasons, the PHYSICAL capabilities of the HARDWARE simply won't allow it. The hardware is rated by DXO's actual measurements, which fall under their Screen DR statistics.

My guess is that the 1D X will still have 11.something stops of real-world HARDWARE DR. The 12.8 stops is only something that might potentially be possible with the right kind of scaling algorithm, and all it will do is allow you to utilize a little more headroom that is normally consumed by noise in a 100% image (however at the tradeoff of detail and resolution...potentially a LOT of detail and resolution, since it required downscaling). The 12.8 stop DR rating tells you about what SOFTWARE can do if it normalizes (bins) noise, but it does not tell you anything about the physical capabilities of the hardware. (Although if 12.8 stops really is the Print DR, it sounds like the QUALITY of the 1D X's noise is really quite good.)

Just to put things in perspective, and maintain a level playing field with consistent arguments: Screen DR tells you about the hardware (and we don't know this yet, probably won't until DXO actually posts the 1D X results on their site.) Print DR tells you about how clever DXO's testing software is and how capable it is at normalizing noise when downscaling.

 

[bdeutsch]

i guess for some the reputation of the DXOmark has just become a bit better. :

;D ;D
I think this is the first time admin put Canon's DXO mark on the front page

Yeah, I thought he said earlier this year that he was never going to post anything dxomark-related on the front page ever again because their sensor scores were meaningless.

Edit:

I know people love to debate the numbers from DXOMark.

Ahh, I guess he's simply realized how great they are for generating page views

I heard a rumor that CanonRumors is posting DXO rumors just to generate page views. Since the source is this forum, I'll rate it CR3, since it has to be true.


Actor Headshots NYC | Gotham Family Photos | NY Wedding Photography

 

[Mooose]

I'm only buying the 1DX if it scores a 91 or better.

 

[Loswr]

Well, the low-light ISO scores are 3296, 2293, and 1815, for the 1D X, 5D3, and 5D2, respectively.

The difference between the 1D X and 5D3 is 1003, and the difference between the 5D3 and 5D2 is 478. So, the difference between the former is ~2x as much as the latter.

Or am I missing something here?

The fact that ISO sensitivity scale progression is not linear, perhaps?

 

[kapanak]

Well, if the numbers have any credence, then the 1D X just became the new low-light king (according to DxOMark), beating the old low-light monster, the Nikon D3s. That's something, right? ;D

Seriously though, who cares about DxOMark? >_>

 

[Marsu42]

Seriously though, who cares about DxOMark?

We do if a Canon beats Nikon ... not so much if it's the other way around.

 

[infared]

DXO has no credibility. I always look elsewhere for credible, thoughtful reviews that are backed up with examples and facts. DXO is a joke.
...hey...I just had a thought...does Ken Rockwell do the testing at DXO??????

 

[jthomson]

...hey...I just had a thought...does Ken Rockwell do the testing at DXO??????

No, Ken thinks the 5D3 is great, despite its low DxO score.

 

[Fishnose]

DXO has no credibility. I always look elsewhere for credible, thoughtful reviews that are backed up with examples and facts. DXO is a joke.
...hey...I just had a thought...does Ken Rockwell do the testing at DXO??????

'DxO has no credibility'.... who says? You and a couple of others here who don't like the results they publish, and just don't understand that DxO CANNOT AFFORD TO BULLSHIT.
Ferchrissake, their test results are just a way to create a buzz around them. They are a software development company who need to test all the bodies and lenses EXACTLY so as to include profiles for them in their software. That's what they do.
And the test results are a byproduct.

Whatever.

Anyway, the 1Dx is an absolutely unbelievably good camera. It's not made to get high scores, it's made to get fantastic results in almost impossible situations, especially low light of course.
And it does that better than any other camera ever constructed.

 

[thepancakeman]

It's not made to get high scores, it's made to get fantastic results in almost impossible situations, especially low light of course.

Silly me; I would think that high scores would be indicative of "fantastic results". In fact maybe that's what some of these folks are saying is the problem with DXO scores.

 

[FunkyCamera]

I guess Nikon's money ran out and they're finally giving proper scores too other sensors.

I wonder if they'll have the courage to go back and put the 5D3 score up to where everyone knows it should be, or if this frightened Nikon enough to get some more Nikon cash in dxo's pockets?

 

[Chewy734]

Well, the low-light ISO scores are 3296, 2293, and 1815, for the 1D X, 5D3, and 5D2, respectively.

The difference between the 1D X and 5D3 is 1003, and the difference between the 5D3 and 5D2 is 478. So, the difference between the former is ~2x as much as the latter.

Or am I missing something here?

The fact that ISO sensitivity scale progression is not linear, perhaps?

yeah yeah yeah Neuro... I just figured it wasn't actually ISO numbers, just some random metric they made up (hence the assumption of a linear scale).

 

[Wrathwilde]

They are a software development company who need to test all the bodies and lenses EXACTLY so as to include profiles for them in their software. That's what they do.
And the test results are a byproduct.

Exactly, I may not agree with their using print DR, or their overall score methodology, but their testing is very exacting, and their raw numbers are probably more accurate than anything else you're likely to find on the web.

If their testing was as bad, or inaccurate as the haters here claim, the DxO Optics Pro software would be a steaming pile of dung that couldn't do anything but degrade the original image. It's not. It's an amazing piece of kit that owes its impressive performance to the fact that DxO's testing is exceedingly accurate. Like it or not, the raw numbers don't lie... although the scoring definitely seems to favor Nikon's strengths.

 

[Tayvin]

Seriously though, who cares about DxOMark?

We do if a Canon beats Nikon ... not so much if it's the other way around.

LOL So true.