DxOMark tests Canon EOS 70D sensor and lenses

In the DPR comparometer you're looking at raw files converted to JPEG using the default settings in Adobe Camera Raw.
DxO's test results, on the other hand, are based on analysis of the unconverted, undemosaiced raw files. They look at numbers, not images. They do this to avoid being dependent on any particular raw conversion software, and come as close to the hardware performance as possible.

And exactly how do they carry out this analysis of the demosaiced RAW files...? By applying their in-house reconstructive interpolation algorithms...? Or do they strip the sensor of its internal circuitry and scrutinize its SNR ratio through external amplification...?

Canon has been using the same 18 MP sensor 4 years back with the 7D. According to DXO mark the DR hasn't improved.

I have processed my friends 7D photos and let me tell you the DR sucked. My A55 was much better and the A55 has got a light grabbing mirror which Nikon bodies lack.

I am no Nikon fan. (Am all at sea with their pro bodies) but facts are facts and Canon sensors just lack the DR performance of Nikon and Sony sensors.

these results are interesting and useful, but they can also be deceiving, especially the combined mark which is a mere sum of the contributions of the characteristics of the sensor. For instance it makes the D800 sound like it completely blows out of the waters the 5D MKIII. Which is somewhat true at base ISO, but if your consistently use relatively high iso then not so anymore. These ratings are just numbers and as such they cannot tell the whole story, however they do give an insight on the potential of the sensor.

don't know if DxOMark test both Nikon and Canon cameras with a same copy of Nikon lens (or a medium format lens for all the SLRs). using multiple lenses we will have to calibrate them carefully with high accuracy.

also shutter speed should be part of camera test. none should we trust, camera ISO, shutter, or aperture (T-number), if we want reliable result which cannot be based on advertised values.

that said, DxOMark is still the best on the web.

I'd be interested to know why people think DXO tests are not BS. What are they based on? Their own made-up methodology? Has anyone put it under scientific scrutiny?

I never really followed the evolution of DXO as a web site and don't really know its origins. I always read this site when Phil Askey owned it and Imaging Resource. I often commented tests on both sites were not scientific in their rigour and I wonder if the same applies to DXO.

DPR, IR and DXO have been around for some time but just because they were fortunate to be the first in on the digital revolution doesn't mean they have the scientific rigour to be held in as high regard as they are. The fact camera companies have to pander to these sites and hope to get good reviews just shows the power of the Internet not that it contains accurate or well derived data.

I am not a Canon user and I am making a general point about all self proclaimed "expert" sites.

^^ If you could post some flaw in their testing methodology that would be great. Why should anyone bother to convince you their testing is flawed ? Rather you should try an convince everyone that their testing is flawed (if at all that is).

"I'd be interested to know why people think DXO tests are not BS."
It's rather the opposite. When choosing to state it is BS, I would seem fit to qualify and quantify that statement.
Accusing, or insinuating, that they are not professional, is ... not professional ... if it is done without anything to back it up.

my comments are based on my own experience with Canon full frame, Nikon full frame, and Pentax 645D. I have no much experience with recent crop sensor technology. I have also seen the eyes of some Nikonians popping when I have shown the highlights recovery from a 5D mK III in Lightroom. Not scientific at all, but based on processing a tonne of images.

can you reference your statement? I am not so convinced with the 85 and 50 mm 1.2 lenses. To me optical and construction quality by Canon is superior

Here you go Amadino.
I used the 6D and the D600 so that you have like cameras in the same price point.
Please note that the 85 1.2 Canon had to be tested at 2.0 for it to perform near the quality of the 85 1.4 at 1.4.
I also included the 85 1.8's from both brands for comparison.

http://www.dxomark.com/index.php/Lenses/Compare-Camera-Lenses/Compare-lenses/(lens1)/940/(lens2)/388/(lens3)/823/(brand1)/Canon/(camera1)/836/(brand2)/Nikkor/(camera2)/834/(brand3)/Nikkor/(camera3)/834

http://www.dxomark.com/index.php/Lenses/Compare-Camera-Lenses/Compare-lenses/(lens1)/823/(brand)/Nikkor/(camera1)/834/(lens2)/241/(brand2)/Canon/(camera2)/795/(lens3)/241/(brand3)/Canon/(camera3)/836

and here is PhotoZone's review of both 85mms...
http://www.photozone.de/nikon_ff/606-nikkorafs8514ff?start=2
http://www.photozone.de/canon_eos_ff/502-canon_85f12ff?start=2

I can look up other reviews for the 35mm's, but needless to say, the results are pretty much the same for the other focal lengths.

Maybe, but in the f/2.8 "section" the new lenses from Canon simply crush everything else. Not speaking about f/4, where the 200-400 has no equivalent on Earth. With a fun fact: for DxO, the Canon EF 70-200 f/2.8 II is "not quite stellar". For the rest of the World (of reviewers, including photozone.de), it is. Cheers! :) P.S.: no worries, Canon will soon start to upgrade the 1.4s... They have no other option, as a jump in sensor resolution is fast approaching.

Dear armandino,
I figured that that test would not be sufficient to prove my point as there are so many DXO detractors, especially considering how poorly Canon does under their testing parameters...

So I also included the lens tests from Photozone which is regarded favorably among DPReview visitors in my original reply.

http://www.photozone.de/nikon_ff/606-nikkorafs8514ff?start=2

http://www.photozone.de/canon_eos_ff/502-canon_85f12ff?start=2

So let me understand,
on your source the 1.4 outscore the 1.2 in shallow depth of field. How seriously should I take that?
Also, I do not doubt that the Nikon is a more all round performer. The Canon lens is designed for more specific applications, I do not think Nikon does better there. The image I posted is an example where I do not think the Nikon 1.4 can do much better. I would be actually surprised if it is as good. I really think your source really misses the point of this lens as much as anybody who has not mastered its personality. And where it shines there is no replacement. Not to mention that Nikon CANNOT even make and 85/1.2.

Ok, so let me get this straight...

You state " optical and construction quality are superior" on the Canon.
I then reference four separate reviews by very reputable sources.
One of which is considered the GOLD standard for methodology, and is often referenced in DPReviews own reviews (DXO).
Showing that performance is better across the board for the Nikon.
You then post one particular image that you shot to show proof of why the 85 1.2 is better, as if this image proves your point.

BTW, the image posted by you is very nice, but by no means is this an example of how this lens outperforms the Nikon since you do not have a side by side comparison shot against the Nikon, unlike the reviewers that used both lenses and came to the conclusion that the Nikon outperforms the Canon at every level....

hmm, my irony-meter fluctuates between 0 and 100% right now...
I would still like to know from Dpr what warrants that comment.

I'm willing to take a bet that it is Shawn Barrnett who has written the text above, could that be correct?

I have mixed felling about this. Everyone wants more DR because it is more forgiving. There are situations where it is actually useful, however I find it can easily lead to HDR effect all the time, overdoing it. Also, I am not a sensor AD conversion engineer, I would be curious if there is any penalty to pay with quantization when stretching the dynamic range with a 14 bit AD conversion.

@Nukunukoo
If you get clipped highlights in studio, then you should adjust the power of the flashes and other light sources. I mean in studio you have the control of the light, so why should you need a high DR?

Noise isn't as easy to process as you think, or how do you get out fine details out of animal fur (or peoples hair) when it is overlaid by noise? And making it look like film grain isn't the solution for some photos.

People who commercially shoot sports also use jpg, which is much more limited in terms of DR, still manage to take good photos for the press or something else without blown out surfaces. And if there are a few shots with clipped highlights, then you still have others.

And for other scenes with movement, you barely use base iso, so the so excellent DR at it won't help you.

@PredatorsPrey
Studio controlled lighting is one thing. Everything else is another. I don't think that needs any further explanation.

Noise have different characteristics and I'm not implying that they are taken nor treated distinctly (nobody does in practice). Noise/DR profiles all play together when processing images. And a good DR helps significantly.

You'd be surprise how good NR is these days or else you would not have questioned it. Of course, ridiculously heavy noise is another thing, as well as banding but we have already developed a technique to reduce that as well.

JPEG output largely depends on the JPEG engine, other better than some. Again, all is not perfect and that's why some come to us. I can, like you, selectively choose situations like sports photography, where they may would use JPEGs for high frame rates. And if you DO take Sports Photography, I don't have to tell you how many shots with bad focus and exposure occur in ratio with the good ones.

Yeah yeah... Ofc, only nikon can make good landscapes... In fact i dont understand why canon users dont give up trying.

Just out of curiosity, do you have any link that shows comparison shots between a Canon sensor and a competitor that demonstrates the gap in DR/noise performance?

Samsung uses Sony sensors.

Samsung phones might but Samsung manufactures its own sensors for its NX range.

Comparing the 70D to the D7000, above 1600 there is essentially no difference in the DR. At 400 there is one stop but why would you even care. If it is DR you want i.e. DR in the sense that DxO measures it (as opposed to what is usually discussed in these forums) why would you EVER set ISO above 100. Once you go above 100, you are trading off DR for something else, in other words, you no longer care about DR.

"If it is DR you want ... why would you EVER set ISO above 100."

Because sometimes the need to stop motion, increase depth of field, and/or minimize camera shake means that ISO 100 won't let you capture the shot! Go shoot a nighttime ball game or live music in a club with available light (sometimes flash isn't allowed and/or produces an undesired result) and see where ISO 100 gets you. In either situation, you might well want more dynamic range than you can get at ISO 3200 or whatever may be needed to get the shot.

Sony NEX-5 -- yes, the original one from 2010 -- 69 DxOMark
Sony NEX-7 -- from 2011 -- 81 DxOMark

Canon has fallen behind Sony on IQ of their sensors (Sony's are used in many of Canon's better competitors, not just Sony bodies). However, I still use my NEX-5 alongside my NEX-7 and I wouldn't reject a 70D based on IQ either. It's the combination of that IQ and a $1200 body-only price tag that bothers me... although I'm sure people will buy it at that price.

D1N0 is a well known Pentax fanboy. The high iso/low light score is NOT that bad on 70D, according dxomark. Canon 70D scores around 920 and K-5 scores around 1220 That's less than 1/2 stop difference.

ET2 is a well known troll. I Have scientific and emperical evidence on my side. You have bullying tactics on your side.