December 20, 2014, 04:46:08 AM

Author Topic: DxOMark Sensor Performance: Nikon vs. Canon  (Read 85878 times)

meli

  • EOS M2
  • ****
  • Posts: 167
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #45 on: September 19, 2012, 06:02:59 PM »
Funny how these threads emerge on this forum with each new Nikon review on DxO.  As an interesting comparison Tech Radar, who is apparantly using the same DxO Analyzer software, recently posted a comparison of SNR and Dynamic Range for the 1DX, 5DIII, 1D4, D4 and D800. http://www.techradar.com/reviews/cameras-and-camcorders/cameras/digital-slrs-hybrids/canon-1dx-1091200/review/page:5#articleContent

While it does show an advantage for the Nikons on base ISO Dynamic Range, the Canons actually shows a better ability to retain it over the ISO range.
:o :P
Really?

actually what techradar results show is that they dont have a clue what they're doing. Just look at the progression of these lines, they dont even make sense!

According to the page you linked we can assume that:

Raw (after conversion to TIFF) dynamic range:
  • D4 has the same DR either @ 100 or 1600
  • 1Dx has peak DR @ 3200 and at 6400 it falls to the same level as @ 100 (magic!)
  • 1DIV has a variation of .5stop through the entire range of isos

JPEG dynamic range:
  • 1Dx peak DR @ 12800  :o
  • 1D4 solid DR from 100 through 4000

Its not the 1st time either that there is a discussion about techradar's silly results, just search around here or any other forum.
« Last Edit: September 19, 2012, 06:06:10 PM by meli »

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #45 on: September 19, 2012, 06:02:59 PM »

sarangiman

  • Canon 6D
  • *****
  • Posts: 375
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #46 on: September 19, 2012, 06:10:29 PM »
Quote
Really?

actually what techradar results show is that they dont have a clue what they're doing. Just look at the progression of this lines, they dont even make sense!

According to the page you linked we can assume that:

Raw (after conversion to TIFF) dynamic range:
  • D4 has the same DR either @ 100 or 1600
  • 1Dx has peak DR @ 3200 and at 6400 it falls to the same level as @ 100 (magic!)
  • 1DIV has a variation of .5stop through the entire range of isos

JPEG dynamic range:
  • 1Dx peak DR @ 12800  :o
  • 1D4 solid DR from 100 through 4000

Its not the 1st time either that there is a discussion about techradar's silly results, just search around here or any other forum.

+1

Ah, yes, I forgot about those results. Pretty strange.

When I mentioned that the general trends between those results & DXO remain the same, I was just pointing to the general DR trends at base ISO (Nikon > Canon), and the general trend that DR for Nikon drops below 1Dx & 5DIII at high ISOs. As well as the trend that pixel-level SNR for D800 falls below that of the Canon sensors -- no surprise there as it has smaller pixels. Normalized SNR, however, is a different story-- with the D800 beating the 5DIII, & the D600 beating both the D800 & 5DIII.

I'm just surprised at how many people call foul at DXO without taking the effort to really analyze what the measurements/results show.

Skulker

  • Canon 6D
  • *****
  • Posts: 410
  • PP is no vice and as shot is no virtue
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #47 on: September 19, 2012, 06:11:45 PM »
.........
Just a little reality check...   :)

Thats not what we want, we want an emotional rant so we can convince ourselves we know it all and Canon are dum and daft.
If you debate with a fool onlookers can find it VERY difficult to tell the difference.

Canon-F1

  • Guest
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #48 on: September 19, 2012, 06:12:32 PM »
Thats not what we want, we want an emotional rant so we can convince ourselves we know it all and Canon are dum and daft.

and sometimes we are right.... kodak anyone?

or sonys TV business... nokia?

a lot of bad decisions can change things pretty fast.
that a company is a market leader today... does not mean it will be tomorrow.
« Last Edit: September 19, 2012, 06:15:59 PM by Canon-F1 »

Skulker

  • Canon 6D
  • *****
  • Posts: 410
  • PP is no vice and as shot is no virtue
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #49 on: September 19, 2012, 06:14:39 PM »
Thats not what we want, we want an emotional rant so we can convince ourselves we know it all and Canon are dum and daft.

and sometimes we are right.... kodak anyone?

Guaranteed we are right, especially with a bit of hindsight.
If you debate with a fool onlookers can find it VERY difficult to tell the difference.

UrbanVoyeur

  • Rebel SL1
  • ***
  • Posts: 80
  • To see and be seen.
    • View Profile
    • UrbanVoyeur.com
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #50 on: September 19, 2012, 06:16:30 PM »
Apologists and bashers. So few realists. Sigh.

The Nikon line using Sony sensors has a consistent, measurable, but modest advantage in DR, color depth and ISO sensitivity over the current crop of Canon sensors at all price points. That cannot be denied. DXOMark and several other sites have shown this.

I'm NOT entirely convinced that these factors alone will result in appreciable difference in image quality under most conditions.

However, when combined with the additional megapixels which DO make difference in cropping and enlargement for publication, the Nikon line has an advantage for now.

I am firmly convinced that if Canon could produce a comparable line of sensors  - same DR, ISO color depth, at the same or higher MP  - for the same price as their current line, they WOULD. They cannot. So this Canon line is the best they can do for right now. Hopefully in the next couple of years they will catch up.

In the meantime, they have been making significant strides in lens quality, so they are not sitting on their hands.

marekjoz

  • 1D Mark IV
  • ******
  • Posts: 944
    • View Profile
    • marekjoz @flickr
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #51 on: September 19, 2012, 06:18:30 PM »
boring...
flickr | youtube | 5D2, 50 F/1.4, 24-105 F/4 L IS, 300 F/4 L IS, x1.4 II

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #51 on: September 19, 2012, 06:18:30 PM »

Canon-F1

  • Guest
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #52 on: September 19, 2012, 06:19:24 PM »
boring...

football?   i 100% agree.... try to take better pictures of a more interesting sport.   ;D

bdunbar79

  • Canon EF 300mm f/2.8L IS II
  • ********
  • Posts: 2606
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #53 on: September 19, 2012, 06:20:43 PM »
We'e to the 6D??  I thought we were still complaining the 1D X was only 18MP.
2 x 1DX
Big Ten, GLIAC, NCAC

Orangutan

  • 1D Mark IV
  • ******
  • Posts: 768
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #54 on: September 19, 2012, 06:26:57 PM »
that does not work very well sometimes.... have a look at sony (tv biz) or nokia for example.

I assume Canon's management and finance people are aware of this.  They're making calculated plans to maximize profit.  Their calculations may be right...or not.  Still, though, it's not personal.  Complaints don't matter, only sales numbers.  Reviews from major photo news outlets have some effect since they probably affect sales numbers.  Complaining on blogs probably does not have a measurable effect.

Besides, I'll bet there are many event/wedding photographers on Nikon sites complaining about the weak AF system in the D800 compared to the 5D3.

Everybody plays with the same laws of physics: I'd be very surprised if Canon does not have high-DR tech in their back pocket.


marekjoz

  • 1D Mark IV
  • ******
  • Posts: 944
    • View Profile
    • marekjoz @flickr
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #55 on: September 19, 2012, 06:33:20 PM »
boring...

football?   i 100% agree.... try to take better pictures of a more interesting sport.   ;D

Looking for some better lenses to do so :)
flickr | youtube | 5D2, 50 F/1.4, 24-105 F/4 L IS, 300 F/4 L IS, x1.4 II

meli

  • EOS M2
  • ****
  • Posts: 167
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #56 on: September 19, 2012, 07:05:20 PM »

I'm just surprised at how many people call foul at DXO without taking the effort to really analyze what the measurements/results show.

+1

confirmation bias, thats why you didnt also get a response over your comment @ dtaylor and his "the right way" DR test.

RuneL

  • EOS M2
  • ****
  • Posts: 201
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #57 on: September 19, 2012, 07:27:57 PM »
That's why I finally decided to switch in next months. I just do want to pay Canon any more. Overpriced , old tech and trust me-very faulty gear, also what is more important seems that they do want to develop any more. I don't wantwaste another two years to get 6D2 with 5 crosstypes and 10EV DR. No more. that's my experience. There nothing much to say. You do not have to agree. Everyone has choice.

 Faulty is not my experience with canon cameras. I've never had a Canon camera fail. I've had ONE problem with the ETTL-circuit in my 1D IV, it was fixed at CPS in 2 hours by changing the top assembly, then only thing it meant was that it couldn't send ETTL info to the flash or shoot HSS, which isn't really a problem for me. I've had them act weird after 4 hours in heavy rain, but they still shot and worked after. My 1D II still works, it has cracks in it where I've dropped it and it's so worn you won't believe it, the batteries are more or less dead, but it still works, not that it ever sees use. I have 16 and 10 year old L lenses that still work fine and are used every week. My 10D and D30 were still alive when I gave them away a few years ago. (I regret getting rid of the D30 though).
The only lens I've ever broken was the 50 1.8 when it was about 4 years old. I treat my gear like S___, I'm not proud of it nor do I do it on purpose, but I'm not good at taking care of stuff I use so much (my cell phones die in 4-6 months) and in such a hurry sometimes. I never use lens caps, I never take care when changing lenses, the mount cap on the lenses may or may not be used, mostly never, because it hinders me as much as a lens cap does, I put my lenses on the ground etc etc etc. And I've never had it fail. I've broken 2 EX 580II's because they've fallen of stands so many times that they finally gave in, but not a manufacturing fault, to my knowledge.

You can be unlucky and get a bad copy, but in my experience they work. Besides my amplifier and speakers, Canon cams and lenses are the bits of electronic gear with the most longevity in my life.

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #57 on: September 19, 2012, 07:27:57 PM »

jthomson

  • EOS M2
  • ****
  • Posts: 176
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #58 on: September 19, 2012, 07:33:27 PM »

=======
96 Nikon D800E
95 Nikon D800
94 Nikon D600
81 Canon 5D III
79 Canon 5D II

(The Canon 1Dx is not yet rated.)

Strange that they can get the Nikon D600 out on the day that it is released but haven't yet got around to the 1DX.

neuroanatomist

  • CR GEEK
  • ************
  • Posts: 15213
    • View Profile
Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #59 on: September 19, 2012, 07:40:36 PM »

IMO, their Measurements (screen) are valid and quite useful.  Their Scores are steaming pile of misleading cow excrement.
Just a little reality check...   :)

(screen) isn't usually the best setting to compare sensor camera vs camera, it can tell you things that are useful, but using (print) to compare various sensors generally makes more sense and for a more fair comparison

I suppose it depends on your goal.  The screen data are the real measurements, print is normalized.  Normalization is fine if you understand what it's doing, but suboptimal if it introduces bias. In this case, while normalizing to a fixed 8 MP purportedly allows comparison across sensors, it also introduces bias - the greater the delta between the sensor MP and the arbitrary 8 MP, the greater the skew - that's part of the reason behind the chart-topping D800 scores.  When analyzing data, one should always be on the lookout for spurious or 'impossible' results - a DR greater that the bit depth of the ADC is an example of that, seen in Print results (which are not quite measurements), but not Screen measurements.

Screen measurements are the raw data, and I always prefer to start with the raw data when doing an analysis. Black boxes are bad science.

which explains how a camera with 14 bits per pixel can, according to DxOMark, actually deliver a dynamic range greater than 14 bits of EV. 

Question for you: my understanding of Bayer pattern sensors is that a group of 4 monochrome {R|G|B} are de-mosaiced to form one color RGB pixel.  If I combine two 14-bit wells I get (roughly) one 15-bit well.  If I combine four 14-bit wells I should get (roughly) one 16-bit well.  Certainly they wouldn't do a crude addition of wells, but it seems very plausible that you can get >14EV DR, even without re-normalizing.  What's wrong with this interpretation?

DxO is measuring the RAW image data, prior to demosaicing, so the ADC is the limit. That's why the Screen measurements are below 14 EV; only the normalized Print values exceed the electronic limit.

I'm just surprised at how many people call foul at DXO without taking the effort to really analyze what the measurements/results show.

I agree, I trust their measurements and in fact, use their software for RAW conversion.  It's their normalized/massaged/weighted-averaged Scores that I don't trust.  I know from direct experience (in my day job) that a set of raw measurements can be reduced using different, internally consistent and seemingly logical methods, to support diametrically opposed conclusions.  I'm very glad that DxO post their raw data, I just don't agree with what they do with those raw data.

Strange that they can get the Nikon D600 out on the day that it is released but haven't yet got around to the 1DX.

That speaks to an entirely different potential bias, but without knowledge about the rationale for the discrepancy, no conclusions can be drawn.  Maybe Nikon overnight shipped them the first D600 off the line, and they're still waiting on a retail back ordered 1D X. Heck, noted bird photographer and Canon Explorer of Light, Art Morris, just recently got his 1D X - I've had mine longer than him.
EOS 1D X, EOS M, and lots of lenses
______________________________
Flickr | TDP Profile/Gear List

canon rumors FORUM

Re: DxOMark Sensor Performance: Nikon vs. Canon
« Reply #59 on: September 19, 2012, 07:40:36 PM »