DXOMark Tests: Nikon D5200 Edges out the D7100?

[Gavjenks]

If people shot film, they wouldn't have to worry about any of them.

? Film has noise, film has diffraction limits, film has MTF charts, film has dynamic range (may vary by developer too but still measurable), etc.

Not sure what you mean here.

If you don't WANT to worry about them though, this is a fantastic website here where you can pull up any cameras and compare their actual images side to side in RAW or JPEG:
Studio shot comparison: Digital Photography Review

As for DxO Mark, Their actual data seem pretty good. Their "overall" and "overall sub" scores or whatever you call them (everything on that page with the picture of the camera and the bar graphs) is pretty much crap, though. I know exactly how they calculate each (they advertise how), and every one of them is dumb. Just as one example, the dynamic range score takes into account ONLY dynamic range at their base rate ISOs, and then completely throws out the other 90% of the data they themselves collected, ignoring it for the score. Why? No particular reason given. This is bafflingly idiotic. Why would they make their score only relevant to people who never shoot anything but ISO 100? The other scores have even weirder issues.

And frankly, due to how incompetently I know they handle their score calculations (something that should be an incredibly simple matter of, for instance, averaging raw measurements over ISOs...), I don't have a lot of faith in their research methods on the other stuff under the hood. Why would they be super careful and reasonable about that, but not the final calculations? Seems unlikely. Their underlying measurements do, admittedly, seem to do a better job than their overall scores at reflecting my actual experience using cameras and looking at real shots, though, so I have not lost all faith in them. Just avoid the scores screens.

Still, the link I just posted above, with the studio shot comparisons that you can just look at yourself, has been gradually earning more of my attention recently.

 

[Derrel]

It's almost ALWAYS Canon shooters who try so,so,so hard to "disprove" and "discredit" DxO Mark scores, or their testing procedures, or some aspect of the laboratory testing and evaluation work that DxO Mark labs publishes...I mean today, now that Canon has lost the lead in sensor technology to Sony-made and Toshiba-made sensors, both of which use newer, more-capable, more-sophisticated .18 micron process sensor fabrication, instead of Canon's outdated .50 micron technology. Back when Canon was in the lead, before the D3 generation was released, the Canon fanboys were CONSTANTLY trumpeting about the fact that Canon cameras had better sensor technology than the other camera makers had. But now that Canon has lost its lead....well...spreading fear, uncertainty, and doubt, FUD, is the new operating procedure for the Canon fanboy types.

I'm pretty sure somebody here is aware of what psychological defense mechanisms are at play in this constant struggle to discredit DxO Mark's sensor test results...

It's amusing....but since I have been shooting d-slr cameras since 2001, and have owned many different models, from three different brands, I'm acutely aware of how ACCURATE and "real-world" the DxO Mark scores actually are. But then again...I bought a $10,000 Canon system to go with my Nikon system, and have actual ownership and actual USE experiences to evaluate DxO Mark data against...so...it might just be my actual ownership and use that makes me understand that, yes Virginia, there actually ARE better and lesser products offered for sale, and the differences are actually real to people who can accept the fact that some sensors are better than others--by a whole bunch at times...OR by "not much at all".

 

[gsgary]

Speaking of these stupid charts, and the Canon owners and users who CONSTANTLY try to "prove" that DxO Mark's measurements have no validity, and have NO real-world implications...I spent a few hours a few weeks back looking very carefully at the DxO Mark scores and graphs for some of the cameras that I actually OWN, and have shot EXTENSIVELY....for literally YEARS with each individual model.

Actually OWNING and having USED the cameras below, I can tell you that the DxO Mark measurements give a pretty damned good real-world indication that shows where the various cameras have real 'issues', and where the newer models are measurably better.

View attachment 55995

And here are the last three Nikon bodies that I bought, and which I still currently own.

View attachment 55996

In each and every case, the DxO Mark figured accurately and truthfully illustrate where each camera is weak. The degrees of difference between the cameras in my comparisons are BIG, and are blatantly obvious, and spending time researching the DxO Mark figures, scree-capping them, and looking at the folder of results has confirmed, and proven to me, beyond a shadow of a doubt, that the DxO Mark results are extremely accurate and have real-world application for people who do not have an agenda to try and "prove". The differences between the D5200 and the D7100 sensor performance are TINY, meaningless differences.

While you wasting your time with those charts i was out shooting and drinking beer

 

[Gavjenks]

It's almost ALWAYS Canon shooters who try so,so,so hard to "disprove" and "discredit" DxO Mark scores, or their testing procedures, or some aspect of the laboratory testing and evaluation work that DxO Mark labs publishes...I mean today, now that Canon has lost the lead in sensor technology to Sony-made and Toshiba-made sensors, both of which use newer, more-capable, more-sophisticated .18 micro process sensor fabrication, instead of Canon's outdated .50 micro technology. Back when Canon was in the lead, the Canon fanboys were CONSTANTLY trumpeting about the fact that Canon cameras had better sensor technology than the other camera makers had.

I'm pretty sure somebody here is aware of what psycholigcal defense mechanisms are at play in this constant struggle to discredit DxO Mark's sensor test results...

It's amusing....but since I have been shooting d-slr cameras since 2001, and have owned many different models, from three different brands, I'm acutely aware of how ACCURATE and real-world the DxO Mark scores actually are. But then again...I bought a $10,000 Canon system to go with my Nikon system, and have actual ownership and actual USE experiences to evaluate DxO Mark data against...

This is not very impressive logic in terms of defending DxO. Even though you're totally right about psychological biases, all of your arguments here ALSO apply just as well to people who do own and/or primarily cameras DxO scores highly. Since there are just as many well known psychological biases in that direction, too. (confirmation bias, cognitive dissonance, etc. apply to both). Owning some of each is helpful, but does not eliminate preconceptions.

Anyway yes, as a Canon owner perhaps I am slightly more motivated to care in the first place about it versus if I shot another brand. But being a Canon owner does not change the fact that it's really really stupid to summarize 10 datapoints by just publishing the first one, for example. No amount of de-biasing will cause that to make any sense. Please, find me any decent scientific peer reviewed articles in any field where they view "listing the first number that we happened to measure" as a valid summary statistic. Median? Sure. Mean? Sure. Mode? Weird in this case, but maybe useful. Range? Sure. Mean, weighted by how often people shoot at each ISO from some surveys? Awesome. "First number on the list?" No.

 

[gsgary]

It's almost ALWAYS Canon shooters who try so,so,so hard to "disprove" and "discredit" DxO Mark scores, or their testing procedures, or some aspect of the laboratory testing and evaluation work that DxO Mark labs publishes...I mean today, now that Canon has lost the lead in sensor technology to Sony-made and Toshiba-made sensors, both of which use newer, more-capable, more-sophisticated .18 micron process sensor fabrication, instead of Canon's outdated .50 micron technology. Back when Canon was in the lead, before the D3 generation was released, the Canon fanboys were CONSTANTLY trumpeting about the fact that Canon cameras had better sensor technology than the other camera makers had. But now that Canon has lost its lead....well...spreading fear, uncertainty, and doubt, FUD, is the new operating procedure for the Canon fanboy types.

I'm pretty sure somebody here is aware of what psychological defense mechanisms are at play in this constant struggle to discredit DxO Mark's sensor test results...

It's amusing....but since I have been shooting d-slr cameras since 2001, and have owned many different models, from three different brands, I'm acutely aware of how ACCURATE and "real-world" the DxO Mark scores actually are. But then again...I bought a $10,000 Canon system to go with my Nikon system, and have actual ownership and actual USE experiences to evaluate DxO Mark data against...so...it might just be my actual ownership and use that makes me understand that, yes Virginia, there actually ARE better and lesser products offered for sale, and the differences are actually real to people who can accept the fact that some sensors are better than others--by a whole bunch at times...OR by "not much at all".

Someone should tell one of the best photographers alive about these charts he uses a really bad performer
Canon Professional Network - Salgado's Genesis project

 

[cbarnard7]

I feel like my "C" grade in undergrad Physics is holding me back from understanding all of this mess...either that or I'm very uninterested.

 

[Derrel]

Someone should tell one of the best photographers alive about these charts he uses a really bad performer
Canon Professional Network - Salgado's Genesis project

Gary...I bought and used the Nikon D2x camera beginning on May 3, 2005. It is a HORRIBLE performer once the ISO levels are moved above about ISO 320...the image quality in multiple metrics declines VERY rapidly...the sensor has VERY bad noise, except at Base ISO level, which is 100. Color richness declines rapidly as the ISO goes up.Dynamic Range drops very quickly too. And yet, despite the sensor's weak performance, at ONE TIME, the D2x was considered a good camera, especially at Base ISO level. Despite the limitations, I made many nice photos with the D2x...poor sensor and all. However, the Canon 5D and the Nikon D3x that I moved to are better, and extraordinarily better, and the DxO Mark figures confirm real-world experience.

The D5200 and D7100 cameras are a virtual DEAD-HEAT, according to DxO Mark sensor test results.

I bought a Canon 5D back in 2007, because, frankly, the image quality the 5D offered was REMARKABLY more-consistent across a wider,more-useful range of ISO levels than what the D2x could deliver...the 5D has pretty reliable, steady imaging performance from base ISO, all the way up to 1600 ISO. and the DxO Mark graph illustrates that...the 5D has a wide-rtanfge, steady performance..the D2x starts off well, but goes to hell, FAST.



Now, the NEWEST full-frame sensor cameras from Canon and Nikon...well, the sensor technology and overall performance has improved to a HUGE degree from what we had a few generations back. And in fact, the better APS-C cameras equal, or out-perform the older-technology full-frame cameras in most respects. Like Gavjenks's brand new Canon 6D...it's got an excellent sensor in it.

 

[gsgary]

I wonder how bad my next digital camera is on those charts, M9 or MM i bet it stinks

 

[Derrel]

Someone should tell one of the best photographers alive about these charts he uses a really bad performer
Canon Professional Network - Salgado's Genesis project

So...Mr. Salgado's old full-frame Canon professional body has image quality that is just a tiny bit better than a Nikon D5200, in some areas, and is equal in color range and tonality, but the newer-sensored Nikon has better dynamic range than his older Canon. Is "that" your point?

"My point" is that a new, APS-C sensor Nikon has equal, or better, sensor performance, than what you call the "really bad performer", the Canon EOS 1Ds Mark III...a formerly $7,995 camera, versus a currently available $600 entry-level Nikon...

 

[gsgary]

Someone should tell one of the best photographers alive about these charts he uses a really bad performer
Canon Professional Network - Salgado's Genesis project

So...Mr. Salgado's old full-frame Canon professional body has image quality that is just a tiny bit better than a Nikon D5200, in some areas, and is equal in color range and tonality, but the newer-sensored Nikon has better dynamic range than his older Canon. Is "that" your point?

"My point" is that a new, APS-C sensor Nikon has equal, or better, sensor performance, than what you call the "really bad performer", the Canon EOS 1Ds Mark III...a formerly $7,995 camera, versus a currently available $600 entry-level Nikon...

View attachment 56004

View attachment 56005

No i'm saying it does not matter how good a camera performs in thoses charts a pro will use what feels best he could have any camera he wanted, people worry too much about high iso and noise

 

[JacaRanda]

It's almost ALWAYS Canon shooters who try so,so,so hard to "disprove" and "discredit" DxO Mark scores, or their testing procedures, or some aspect of the laboratory testing and evaluation work that DxO Mark labs publishes...I mean today, now that Canon has lost the lead in sensor technology to Sony-made and Toshiba-made sensors, both of which use newer, more-capable, more-sophisticated .18 micron process sensor fabrication, instead of Canon's outdated .50 micron technology. Back when Canon was in the lead, before the D3 generation was released, the Canon fanboys were CONSTANTLY trumpeting about the fact that Canon cameras had better sensor technology than the other camera makers had. But now that Canon has lost its lead....well...spreading fear, uncertainty, and doubt, FUD, is the new operating procedure for the Canon fanboy types.

I'm pretty sure somebody here is aware of what psychological defense mechanisms are at play in this constant struggle to discredit DxO Mark's sensor test results...

It's amusing....but since I have been shooting d-slr cameras since 2001, and have owned many different models, from three different brands, I'm acutely aware of how ACCURATE and "real-world" the DxO Mark scores actually are. But then again...I bought a $10,000 Canon system to go with my Nikon system, and have actual ownership and actual USE experiences to evaluate DxO Mark data against...so...it might just be my actual ownership and use that makes me understand that, yes Virginia, there actually ARE better and lesser products offered for sale, and the differences are actually real to people who can accept the fact that some sensors are better than others--by a whole bunch at times...OR by "not much at all".

Someone should tell one of the best photographers alive about these charts he uses a really bad performer
Canon Professional Network - Salgado's Genesis project

Wow, I have 3 of the same lenses and the extender. I feel less of an idiot!

 

[Derrel]

SNIP>> people worry too much about high iso and noise

Well, yes and no. Or alternately, no, and yes! I grew up and learned to shoot on B&W film of ISO 125, Plus-X Pan. High-speed B&W film was 400 ASA Tri-X, and 400 ASA is pretty useful. It was, I believe, the summer of 1977 or '78 when 400 ASA color print film was invented. ASA 400 or Kodak High Speed Ektachrome looked pretty grainy, and was expensive.

But what has happened in the last five years is that top-level d-slr image quality has gone wayyyyyyyyyyyy, wayyyyyyyy up at very high ISO levels, like 6,400, and 12,800, and 25,600, and so on; a few years ago, ISO values above 3200 were utter rubbish, but today, the newer d-slr cameras offer MUCH better quality than ANY film can. So, to people who want to use smaller lenses, or shoot in lower light, or push the boundaries of what is possible, the NEWER d-slrs have expanded the range of what is possible.

But your example is interesting: a cheap Nikon crop-body now produces basically, about the same technical image quality as a formerly $8,000 Canon full-frame body. What the newer High-ISO capable cameras have done is made "f/4" lenses "the new f/1.4" lenses...and that is why we're seeing more new pro-grade f/4 zooms, like the new ones from Canon and Nikon. Canon's new 24-70mm f/4 L IS USM lens and Nikon's new 16-35mm f/4 VR-II lens are good examples of pro-grade lenses that are now being made because High ISO performance has improved so,so remarkably as of late.

 

[AutofocusRoss]

Someone should tell one of the best photographers alive about these charts he uses a really bad performer
Canon Professional Network - Salgado's Genesis project

So...Mr. Salgado's old full-frame Canon professional body has image quality that is just a tiny bit better than a Nikon D5200, in some areas, and is equal in color range and tonality, but the newer-sensored Nikon has better dynamic range than his older Canon. Is "that" your point?

"My point" is that a new, APS-C sensor Nikon has equal, or better, sensor performance, than what you call the "really bad performer", the Canon EOS 1Ds Mark III...a formerly $7,995 camera, versus a currently available $600 entry-level Nikon...

View attachment 56004

View attachment 56005

No i'm saying it does not matter how good a camera performs in thoses charts a pro will use what feels best he could have any camera he wanted, people worry too much about high iso and noise

You hit the nail on the head. Put an idiot in a Ferrari, what do you get, a bent Ferrari. Give an idiot a D7100 / D800 what do you get... well, it begins with c, ends in p, and there are two letters missing. However, give a genius a D3100 and you'll get stuff worthy of the National Geographic Magazine. That's a sad fact, the camera is only a tool to help you achieve something wonderful - when the camera itself becomes the most important thing, we are in danger of losing ourselves and our hobby.

That said, I am still glad I waited to see all the reviews of both of these cameras before getting the 5200. Now I can relax and enjoy learing all it's little tricks and nuances... a 240 page manual is worrying, fortunately my 30 years of using Av on SLR's puts me in good stead, and no reciprocity failure to worry about... or Kodachrome for that matter.

Happy days!

 

[Derrel]

Anyway yes, as a Canon owner perhaps I am slightly more motivated to care in the first place about it versus if I shot another brand. But being a Canon owner does not change the fact that it's really really stupid to summarize 10 datapoints by just publishing the first one, for example. No amount of de-biasing will cause that to make any sense. Please, find me any decent scientific peer reviewed articles in any field where they view "listing the first number that we happened to measure" as a valid summary statistic. Median? Sure. Mean? Sure. Mode? Weird in this case, but maybe useful. Range? Sure. Mean, weighted by how often people shoot at each ISO from some surveys? Awesome. "First number on the list?" No.

Um, dude, are you even aware that there are breakouts, graphing the performance across the ENTIRE ISO range of the tested cameras? (The Nikon D70 for example, has a range from 200 to 1600 ISO, no more at all, no ISO expansion,etc..)

Once again, your disingenuous attempts to slant the facts is showing...

As one can easily see, DxO Mark shows a whole RANGE of data points. Cough,cough,cough. Oopsies!



Pretty much the ENTIRE ISO range of all tested cameras is tested and graphed out. But, then hey, who's paying attention to the facts? Your attempt to try and discredit the validity and or utility or usefulness the test results surely cannot be because you're unaware that the WHOLE data spectrum *is actually available*..I mean, you're smarter than that,right? And yet...you conveniently forget to note that the ENTIRE set of data points is *available* with just a mouse-click or two! Hmmm....I doubt your motives, seriously, I do.

 

[Gavjenks]

Anyway yes, as a Canon owner perhaps I am slightly more motivated to care in the first place about it versus if I shot another brand. But being a Canon owner does not change the fact that it's really really stupid to summarize 10 datapoints by just publishing the first one, for example. No amount of de-biasing will cause that to make any sense. Please, find me any decent scientific peer reviewed articles in any field where they view "listing the first number that we happened to measure" as a valid summary statistic. Median? Sure. Mean? Sure. Mode? Weird in this case, but maybe useful. Range? Sure. Mean, weighted by how often people shoot at each ISO from some surveys? Awesome. "First number on the list?" No.

Um, dude, are you even aware that there are breakouts, graphing the performance across the ENTIRE ISO range of the tested cameras? (The Nikon D70 for example, has a range from 200 to 1600 ISO, no more at all, no ISO expansion,etc..)

Once again, your disingenuous attempts to slant the facts is showing...

As one can easily see, DxO Mark shows a whole RANGE of data points. Cough,cough,cough. Oopsies!

View attachment 56007

Pretty much the ENTIRE ISO range of all tested cameras is tested and graphed out. But, then hey, who's paying attention to the facts? Your attempt to try and discredit the validity and or utility or usefulness the test results surely cannot be because you're unaware that the WHOLE data spectrum *is actually available*..I mean, you're smarter than that,right? And yet...you conveniently forget to note that the ENTIRE set of data points is *available* with just a mouse-click or two! Hmmm....I doubt your motives, seriously, I do.

Considering that A) I mentioned these above in this thread already, and B) you mainly began posting the breakout graphs after I complained about you not doing it multiple times, yes, I'd say it's safe to say that I am aware of those graphs.

And like I said, these raw measurements do indeed seem to do a better job of matching real experience than the scores. If you're going to cite DxO, you should definitely cite the graphs, as you have been doing.




HOWEVER,

A company's reputation is a holistic thing. Why should I believe that they acted like complete idiots in their summary statistics, and yet somehow acted like wise, careful professionals in everything else they do for gathering the raw data itself? If somebody does one set of things that are really dumb, it reflects on their professionalism and knowledge and level of laziness in general, across ALL areas.

That single series of multiple major statistics errors very reasonably casts doubt on the scientific skills of the ENTIRE organization and EVERYTHING else that they do. Especially since it is the main set of numbers that they advertise by default to anybody on their site.

In other words, I am assuming that they acted similarly carelessly in their raw data collection. The fact that the raw data is more accurate seeming could merely be a consequence of fewer errors having piled up on top of one another than in the overall scores.