DXO: 7D vs. D5000

[Derrel]

Wow, according to DXO the D5000 even trumps the more expensive D300s.

I smell something fishy.

I would say it's more a case of the D5000 produces insignificantly different quality results out of the camera than the D300s. This is one very small aspect of the camera though given the D5000 lacks many of the features like the awesome autofocus system.

I highly doubt this site would measure RAW performance of the sensor.

They measure the sensor based ONLY upon its RAW data. DxO Mark uses NO JPEG information in its evaluation. Note the pixel pitch difference between the D5000 and D300s--two different pixel pitches. the 7D has the smallest pixels of any d-slr on the market....smallest pixels of any d-slr ever made actually. I'm not sure there's another gunman on the grassy knoll, but we could ask Oliver Stone to see if he can find anything fishy.

 

[inTempus]

It would appear, according to DXO Labs, that Canon hasn't improved upon the 20D with any of its subsequent releases. Here we can see the 20D matches the performance of the 50D and even the new 7D. Well, the 7D apparently just edges it out in ISO performance whereas the 50D can't beat the 20D.

I also find it suspicious that the 7D beats the D300s in ISO performance given it is 18mp with a notably smaller pixel pitch and the D300s is 12mp.

 

[Derrel]

It would appear, according to DXO Labs, that Canon hasn't improved upon the 20D with any of its subsequent releases. Here we can see the 20D matches the performance of the 50D and even the new 7D. Well, the 7D apparently just edges it out in ISO performance whereas the 50D can't beat the 20D.

I also find it suspicious that the 7D beats the D300s in ISO performance given it is 18mp with a notably smaller pixel pitch and the D300s is 12mp.

Actually, your statement is incorrect; the 7D has a wider dynamic range than the 20D, by 2/3 of an f/stop, which is actually a pretty substantial improvement over that of the 20D. Your statement is incorrect too, in that the low light ISO of the 7D is improved over the 20D. Your statement is also incorrect in that the 7D's overall sensor score is higher than Canon's 20D. Canon HAS made improvements since the 20D--while pushing boundaries!

As to their findings that the 7D beats the D300s in ISO performance with its 18MP sensor and smaller pixel pitch; a d-slr is more than "just" a sensor--there are many,many other factors. One of the easiest ways to reduce noise coming off the chip is to increase the read-out channels. The sensor itself is only one part of the equation. The 7D is newer, and it is Canon; the D300s is an incremental "s" update (Nikon's equivalent to Canon's MK or (n) nomenclature and is basically a late 2009 update (tweak) to a 2007 camera with a 2006 design--it is almost 2010 now. I would expect that the differences between the two cameras are pretty substantial in how the two companies process the sensor information.

Another thing, and this is not to be underestimated, is how GOOD the 20D actually WAS. I have owned a 20D since it was new, and still shoot it. Until the EOS 1D Mark III hit the market, Rob Galbraith proclaimed on his website that the JPEG images coming out of the 20D were the cleanest, most-noise-free, and all around BEST JPEG images out of any camera they had used. An that included professional Canon and Nikon bodies! Even though the 20D is several generations old now, it has the 6.3 micron pixel pitch, it has only 8.2 million pixels, AND its ISO can be tested only in its native range which is 100-1600; the 20D does fantastic work up to 800, 1600 is quite good, but has some noise. The 50D and 7D are tested over the range of ISO 100 to 12,800--which is a 3-stop or 8-fold difference in ISO range. Although the 7D shows some improvement over the 20D in low-light ISO, with the older 20 pulling a score of 721, and the 50D only 696, and the 7D coming in at 809, the two newer cameras have the DISADVANTAGE of the ISO 12,800 performance, as well as the 6400 and 3200 ISO performance on which they are also evaluated. And, let's face it--the 50D and 7D are really,truly "pushing the envelope" of ultra-small pixels on a smallish sensor with ridiculously high ISO ranges on which they are evaluated. I think the 7D's ultra-high ISO scores at 12,800 and to a slightly lesser extent the ISO 6,4000 scores are pulling the 7D's average wayyyy down.

Color depth has not improved much over the models, but it was already good to begin with, and to get better depth you need to move to a much more expensive class of camera. Dynamic range HAS improved, and I think that is largely due to better signal processing, and better microlenses and in the 7D.

No offense is meant, but I think your selection of cameras is a self-serving one; why not compare the advances Cano has made in the full-frame arena against the 7D? You will see,clearly, that low-light ISO is the domain of the FF sensor cameras from Nikon and Canon.
Just for example--the original Canon 5D; like the 20D, it was a benchmark camera for its time. The original Canon 5D has huge pixels, 12.8 MP sensor size,and has a really GOOD image quality. At lower ISO levels, like from 100 to 800, the old 5D is about as good as the Nikon D3--maybe a bit better on a per-pixel level according to my eyes.
And one other thing about the DxO Mark system that I happened to read on Luminous Landscape--they do NOT consider resolution in their evaluation. So, the higher MP counts of the 50D and 7D over the 20D are really not counted. I simply cannot agree with your statement that Canon has "not improved upon the 20D with any of its subsequent releases," since the data shows there is improved dynamic range and improved low light ISO performance, plus significantly higher resolution. AND they have managed to boost the ISO range by a factor of EIGHT TIMES while making these improvements. I think if the 7D topped out at ISO 1600,like the 20D did, the low-light ISO improvement score would be higher, but then again, Canon has pushed very,very,very far--the 7D has the smallest pixels of ANY d-slr on the market AND it has 12,800 ISO---that's something that was unheard of not too long ago. And, if you compare the 7D with the 5D Mark II or the Nikon D3s, or even the "old" 5D, you'll see that the real improvement has been in the FF sensor cameras. The 20D was a sweet spot camera for Canon; measure its low light ISO against the $5,000 Nikon D2x for example.

I'd like to see a screen shot comparison of the 7D with the Canon 5D and the Nikon D3s and the 5D Mark II--each leaders in their respective classes.

 

[inTempus]

Let me rephrase, the differences between the 20D, 50D and 7D is pretty small, and you seemed to focus on the performance of the 7D and not so much on the 50D.

The differences I would expect to see in sensor performance between 4 generations of bodies would be something like this:

I don't need to show a 5D2 vs. a D3s because that's not what I was talking about. I'm talking about advances made in a product line, not comparing Nikon to Canon (yet again). I fail to see what's "self serving" about this comparison.

I would be interested to know how the 20D out performs the 50D in ISO performance and how that specifically is measured.

***Edit: The D2X wasn't full frame. I always forget that.

 

[inTempus]

Humm, here's the D200 vs. the D300s. It seems Nikon improved their bodies a little more, but it's on par with the differences between the 20D and the 7D (not so much the 20D vs. the 50D). So Nikon has improved their bodies in 1 or 2 generations a little more than Canon has in 4.

 

[Derrel]

Let me rephrase, the differences between the 20D, 50D and 7D is pretty small, and you seemed to focus on the performance of the 7D and not so much on the 50D.

The differences I would expect to see in sensor performance between 4 generations of bodies would be something like this:

I don't need to show a 5D2 vs. a D3s because that's not what I was talking about. I'm talking about advances made in a product line, not comparing Nikon to Canon (yet again). I fail to see what's "self serving" about this comparison.

I would be interested to know how the 20D out performs the 50D in ISO performance and how that specifically is measured.

***Edit: The D2X wasn't full frame. I always forget that.

Sorry, but you've missed most of my points. The 20D was a BENCHMARK camera, the best of its type and era. Its low-light performance SMOKES the Nikon D2x's low-light high ISO performance. Also, the 20D-30D-40D-50D-7D
isn't really four generations as much as four really rapid "iterations" ; the 20D and 30D shared the same sensor at 8.2MP, the 40D went to 10.2, and the 50D was basically the 40D souped up. Canon iterated at 18-month intervals, as fast as they could; Nikon goes much more slowly, with 4-year intervals between flagship models like D2x to D3x,and the D2x-D3x comparison is again, APS-C vs FF.

My point has been proven by your gracious screen capture--which is not self-serving because somebody other than you, somebody other than YOURSELF, asked you to consider another point of reference. I mentioned the D2x so you could get a comparison about how superb the 20D's sensor was for its time--even compared to a $5,000 professional "flagship" Nikon.

The 7D and the 50D are both pushing the envelope. The 20D topped out at ISO 1600; the 50D and 7D top out at 12,800-which is an EIGHT-fold increase! That is shown in their data. That is an improvement.

Your comparison was chosen by you, yourself, hence the use of the word self-serving,and your conclusion was in direct contradiction with the figures; you said no improvement, but there was some improvement AND huge added new capabilities you failed to mention-resolution, frame rate, Dynamic range, overall scores,and all at the same price.

Do you fail to see how awesome the 20D was in comparison to the Nikon D2x--the 20D smokes the D2x on overall sensor performance score, dynamic range, and low-light-ISO score! The D2x and 20D were contemporaries. As I pointed out, the Canon 20D was a BENCHMARK camera. A benchmark is something by which OTHER things are measured. I do not throw the word benchmark around lightly: the $1700 EOS 20D SMOKED the $5,000 Nikon D2x camera in terms of SENSOR performance. How do I know that???? Well, I OWN both cameras, the 20D and the D2x,and have for several years. I have shot both extensively, at all ISOs,and under all types of lighting. And you know what? The EOS 5D has a better sensor than either of those two cameras.

Rest assured, the D2x's low-light ISO score of 467 is well-deserved; the new EOS 7D is almost TWICE as good. The original 20D, which you selected for the point of initial comparison was, I will say it again, a BENCHMARK camera. Larger pixels than the D2x, lower pixel density, a refined CMOS sensor Canon had been working on for some time. The D2x had a one-off sensor, the only CMOS sensor of its type that Nikon ever used. Your point of initial comparison was a SUPERB product!

You picked one of Canon's finest cameras, the 20D, as a starting point, and concluded that they had made "no improvements", when in fact, Canon has made MAJOR improvements in the 7D's capabilities and the data show the improvements.

As an owner of the 20D and the D2x I will state flat out: the DXO Mark scores for the D2x and the 20D are pretty much reflective of how the cameras SENSOR's perform. I own a 20D and a D2x,and can speak from actual experience as to how relevant and accurate the DXO Mark scores are WRT to those two bodies. I also own the Canon 5D, which I also suggested you do a screen cap of as a comparison, since the 5D is also a true "benchmark" camera.

The 20D was and still is a very capable camera. How come the new 2009 Corvette isn't 75 miles an hour faster than the 2005 model? I'm not sure how much more can be done with pixels of 4.3 micron size and with ISO 6400 and ISO 12,800--can Canon engineers walk on water?

 

[inTempus]

Are you such a Canon fanboy that you cannot read the word Nikon and fly into a rage?

You're the only one who has flown into a rage... again. Seems to be a trend with you. Rage and name calling... I'm curious, where's the comments about my signature line, user name, website, an any other off-topic taunt you can think up?

I try to have a civil discussion with you and you immediately resort to name calling when I question a comment you've made or post an opposing view.

Until you learn how to be civil, I'm done responding to you in this thread.

 

[JamesMason]

Whilst im aware that the 7d for many reasons is a much much better camera.

DAMN THATS FUNNY !!!

 

[Dao]

I really do not pay much attention on DxOmark. Especially after I read this article.

Eyes vs. Numbers

 

[inTempus]

I really do not pay much attention on DxOmark. Especially after I read this article.

Eyes vs. Numbers

Interesting read. I found this particularly interesting:

Does a ranking of 62.3 really differ in any meaningful way from 63.8? No, not at all. In fact DxO points out that a measure smaller than 5 is hardly perceptible, representing just a 1/3rd stop difference.

Not ".5" but "5". That means if one camera rates say 22.1 on Color Depth and another rates at 23.6 the difference isn't really anything perceivable by the human eye.

 

[benhasajeep]

That artice has some very good points. And one that I alluded too. And that was the 7D performance was very close to the others but at 18mpix. I thought that was pretty impressive. Yes the others would have a better picture (did not know it would be as small of a difference as it really is). But the output of the 7D would outshine the others once you started to get into bigger prints. Going by that article, they have no merits for output size. In reality if a company made a very high dynamic range 1mpix full frame camera that has extremely low noise at ISO 25,600 or higher. It would score as the best camera made! And yet it would only produce a very small picture. I think thats a pretty big flaw in the system.

 

[Derrel]

Dude, I'm not raging. I'm just going over all this data you have elected to bring to the forum.

Look--the 7D has made major improvements over the 20D. I explained it all very carefully above, as I see it.

I'm not name-calling; you're the one who brought up DXO Mark's credibility; all I can go by is how well their scores match with my experience. You say their data is "fishy"....but it seems reliable to me.

From my point of view, the 5D, 20D, D2x, D40 and D70 and Fuji S5 Pro scores given by DXO Mark ALL reflect my own person experience with those actual cameras. If you want to cast aspersions on either Canon or DXO Mark, by all means, let 'em have it. I think Canon has done a good job with the 7D.

I consider an eight-fold ISO increase and a bump from 8.2 to 18MP and a dynamic range improvement of 2/3 of a stop to be HUGE improvements over the 20D's originally stellar performance. And at eight frames per second, body-only,with no grip needed? And all for $1699? What more can Canon do? The 7D is the highest-resolution,fastest-firing, best-video-in-class,sleekest,most feature-laden camera in the APS-C class. But like the 2005 Corvette and the 2009 model, the "engine" (ie sensor) performance has not undergone a quantum leap--Canon is too far along for those kinds of improvements now.

 

[inTempus]

Dude, I'm not raging. I'm just going over all this data you have elected to bring to the forum.

I have no problem discussing the data. It's when I am called a fanboy that things go off track and take the discussion from a civil discourse to another unwarranted and unwelcome flame war. I would like to avoid that if we can.

I also find it a bit perplexing as to why you continue to call me a fanboy even when I say Nikon has done something better than Canon (like in this thread). I am far from a fanboy of Canon, and I often times make negative comments about various products or services they offer (as well as positive). I have no allegiance to any particular brand. Now, if someone says something I feel is inaccurate about Canon (or any other brand), I will point that out. That doesn't make me a fanboy.

Look--the 7D has made major improvements over the 20D. I explained it all very carefully above, as I see it.

I would agree there are features that are major improvements, but looking at the numbers and the reading this article from Luminous Landscape, it would appear the improvements to IQ aren't all that major. They're actually barely perceivable if that article is to be taken as factual. You've quoted articles from them in the past so I must assume you find their opinions to be generally accurate.

I'm not name-calling; you're the one who brought up DXO Mark's credibility; all I can go by is how well their scores match with my experience. You say their data is "fishy"....but it seems reliable to me.

I would disagree in that calling me a "fanboy" isn't name calling. I believe it is. I think we would get along much better if we avoided such insults going forward. My questioning DXO's accuracy isn't an insult to you in the slightest, so calling me a fanboy in response to my inquiring into their accuracy isn't warranted.

From my point of view, the 5D, 20D, D2x, D40 and D70 and Fuji S5 Pro scores given by DXO Mark ALL reflect my own person experience with those actual cameras. If you want to cast aspersions on either Canon or DXO Mark, by all means, let 'em have it. I think Canon has done a good job with the 7D.

I'm glad you think the 7D is a great camera. I was cold on the camera initially given the specifications, but after having played with one a few times now I'm pretty impressed with it. I do believe overall it's a vast improvement over previous xxD models (even though it's in the xD group).

I consider an eight-fold ISO increase and a bump from 8.2 to 18MP and a dynamic range improvement of 2/3 of a stop to be HUGE improvements over the 20D's originally stellar performance. And at eight frames per second, body-only,with no grip needed? And all for $1699? What more can Canon do? The 7D is the highest-resolution,fastest-firing, best-video-in-class,sleekest,most feature-laden camera in the APS-C class. But like the 2005 Corvette and the 2009 model, the "engine" (ie sensor) performance has not undergone a quantum leap--Canon is too far along for those kinds of improvements now.

The ISO improvements are notable. My comments about the lack of any notable improvements were directed mostly at the Dynamic Range, Color Depth and overall Sensor ratings.

You won't get an argument from me, the 7D is a great camera. My only point in all of this is that the measurements offered by DXO seemed odd (especially when comparing the D300s to the 7D). Lots of other people around the internet find their numbers to be odd at times... I dare to say most people I've read comments from generally frown upon them. That doesn't make them inaccurate, but it is interesting they're so maligned. Perhaps it's due to brand partisanism. Who knows.

But I do want to make an effort to get along with you Derrel. I don't like battling with you in every thread. Know that if I question something you say, it's only because I'm looking for more information. If I'm wrong, I'll admit it. But I don't view many of our conversations as "right and wrong" as much as I seem them as an exchange of thoughts and a learning experience.

 

[cfusionpm]

According to DXO Marks, the 5D mkII and Nikon D700 rank higher than a Hasselblad H3DII. Fishy enough?

Quotes from the link "Eyes vs Numbers:
"This scale is based on three underlying metrics, Color Depth, Dynamic Range and Low-Light ISO.
The above quote from the DxOMark web site indicates the nature of the problem. Sensor resolution is not one of the metrics included. This means that a 12MP camera that scores a couple of points higher than a 24MP camera will be judged by most people as being superior by its DxOMark rating, even though, as we now know, less then 5 points difference is not even visible. Yet, camera one has twice the pixel count (1.4X the resolution) of the other. This flies in the face of both experience and common sense. "​

"I am not disputing DxO's numeric test results. They show what they show. But the numbers themselves simply do not correlate with the reality that I and many other knowledgeable photographers see, and which literally thousands of professionals around the world experience in their work on a daily basis."

That entire link pretty much nullifies any real world application of these numbers. Sure, you can try to use them to justify purchases or say your camera is better than mine, etc, etc. But because of their methodology and omission of important elements, they do not reflect real world results.

 

[Sw1tchFX]

i wonder if they use the Hasselblad in the 4-shot mode. in that, the color depth should blow pretty much anything out of the water.

As far as noise though, MF backs are terrible. We've done long exposures in the studio for cars on an H2D22, and after about 20 seconds, the tread on the tires disappear from the camera doing noise reductoin. A 5D or D700 would easily have more detail and print better.