DxOMark scores for 5DMkIII out - total score 81, 5DMkII had 79

[ctmike]

The numbers are the numbers; the approach seems reasonable... for the specific conditions they test. It's still related to the sensor only, and it doesn't take away from the generally awesome performance of the MkIII.

Canon does have a significant marketing problem on their hands now, though. People are going to read today and for the foreseeable future that : Nikon Sensor > Canon Sensor, Nikon 36MP > Canon 24MP, and D800 costs less than MkIII. Most people aren't going to get in to many philosophical debates about what the DxOMark scores actually mean, they are going to see the above and buy Nikon. Quite frankly, Canon is going to have this problem at the low end as well, with whatever they release to compete against the D3200.

I love Canon, but they got caught with their pants down, marketing-wise. Doesn't mean that their cameras take bad pictures. But taking great pictures is only part of the story when people decide what to buy.

 

[kozakm]

Re: DX0 Mark Canon 5D MkIII Review

The only test i've seen banding and color blotches in, they took a shot underexposed, then added 4-5 stops of exposure to show the banding & blotches. I'm fine in most cases with pixel peeping, but seriously, there is no real world example to when you would need to push the exposure that far. Those kinds of tests are rubbish.

Not true. You definitely don't need to add 4 to 5 stops. And real world example? Shooting at the bright sunlight with deep shadows. I also tried several Nikon raw files and din't notice behavior like this.

 

[skitron]

Personally, I think DXO provides a valuable service in posting their findings.

That said, two things: first, don't expect anything meaningful out of a single number (i.e. DXO mark) and second, when you DO look at the details, it's a good idea to keep things in perspective (i.e. the scale of the graphs matters):

http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/795%7C0/(brand)/Canon/(appareil2)/792%7C0/(brand2)/Nikon/(appareil3)/767%7C0/(brand3)/Nikon

As we can see from the Measurements:

1) ISO sesitivity - no meaningful differences except the total range (5D3 beats D800 and D4 beats 5D3)

2) SNR 18% - no meaningful difference except high ISO (5D3 beats D800 and D4 beats 5D3)

3) Dynamic range - some meaningfull differences where D4 wins it all, D800 beats 5D3 below ISO 2400, 5D3 beats D800 above ISO 2400

4) Tonal range - some slightly meaningfull differences where D4 wins it all, D800 seems to beat 5D3 below ISO 2400, 5D3 seems to beat D800 above ISO 2400 and of course 5D3 extends the total range over D800

5) Color sensitivity - some meaningfull differences where D4 loses to D800 below ISO 360 and 5D3 beats D800 above ISO 16000

Bottom line is three fantastic cameras, just take your pick and be happy!

 

[AJ]

well, here, how I look at it.

The camera: it is what it is.

The dxo score: hopefully it'll put some pressure on for Canon to lower the price, sooner rather than later.

 

[Fishnose]

I love Canon, but they got caught with their pants down, marketing-wise.

Yes, success leads to complacency.
Just like the big trouble Nokia is in now. They had 37% of the whole cellphone market a couple of years ago, now they're #2 after Samsung and losing speed fast.

Situation:
2 Canon guys, from R&D and Sales, are sitting out in the sun, taking a (really long) break.
Something yellow and black comes by at tremendous speed, so fast they can't really tell what it was.
"A bumblebee, maybe?" says one.
"Yeah, I guess."
And they go on relaxing.

What was it?
Nikon, going really fast.

 

[Loswr]

1) ISO sesitivity - no meaningful differences except the total range (5D3 beats D800 and D4 beats 5D3)

Just to call this one out, DxOMark's ISO Sensitivity measurement isn't measuring what most people think it's measuring. It has no direct bearing on ISO noise performance (but an indirect one, see below). The name is perhaps misleading, maybe better to call it 'ISO Accuracy' or 'ISO Fidelity'. What this test measures is the real ISO value (benchmarked against the actual International Organization for Standardization's criteria) vs. the ISO setting on the camera, or to put it another way, it measures how much the camera lies to you when you pick a given ISO setting.

In the plot you can see that for ISO 50 and ISO 100, the dots for the 5DIII, D800, and D4 are all stacked on top of each other, and they're all at ISO 75 for both settings. What that means is all three cameras are lying to you in exactly the same way - whether you set ISO 50 or ISO 100 for your shot, the exposure is actually at around ISO 75 and then pushed or pulled by the camera as needed, although the ISO value you selected is what's actually recorded in the metadata. This lying is not new or unique - both Canon and Nikon do it routinely for fast lenses, where the incident angle of the light exceeds the refracting capability of the microlenses and exposure is 'secretly' boosted to compensate (i.e. about 1/2 stop of the light coming in at f/1.2-1.4 is not detected by a digital sensor, so the camera boosts the ISO half a stop - meaning half a stop more noise - without telling you).

Perhaps of a bit more significance is the way this plays out - if you compare just the D800 to the 5DIII and look at the ISO Sensitivity plot, you can see that the D800 is a little further off the nominal value at all the settings, with more separation at the higher ISOs. In other words, the D800 lies to you a little more than the 5DIII (but the 5DIII is still lying). For example, when you set both cameras to ISO 6400, the D800 is actually shooting at ISO 4211 (it's lying by 2/3-stop), whereas the 5DIII is actually shooting at ISO 5179 (it's lying by only 1/3-stop). Translation - artificial advantage for the D800 because it's shooting at a lower actual ISO than the 5DIII for a given setting.

Bottom line is three fantastic cameras, just take your pick and be happy!

This is the take home message!

 

[skitron]

1) ISO sesitivity - no meaningful differences except the total range (5D3 beats D800 and D4 beats 5D3)

Just to call this one out, DxOMark's ISO Sensitivity measurement isn't measuring what most people think it's measuring. It has no direct bearing on ISO noise performance (but an indirect one, see below). The name is perhaps misleading, maybe better to call it 'ISO Accuracy' or 'ISO Fidelity'. What this test measures is the real ISO value (benchmarked against the actual International Organization for Standardization's criteria) vs. the ISO setting on the camera, or to put it another way, it measures how much the camera lies to you when you pick a given ISO setting.

In the plot you can see that for ISO 50 and ISO 100, the dots for the 5DIII, D800, and D4 are all stacked on top of each other, and they're all at ISO 75 for both settings. What that means is all three cameras are lying to you in exactly the same way - whether you set ISO 50 or ISO 100 for your shot, the exposure is actually at around ISO 75 and then pushed or pulled by the camera as needed, although the ISO value you selected is what's actually recorded in the metadata. This lying is not new or unique - both Canon and Nikon do it routinely for fast lenses, where the incident angle of the light exceeds the refracting capability of the microlenses and exposure is 'secretly' boosted to compensate (i.e. about 1/2 stop of the light coming in at f/1/2-1.4 is not detected by a digital sensor, so the camera boosts the ISO half a stop - meaning half a stop more noise - without telling you).

Perhaps of a bit more significance is the way this plays out - if you compare just the D800 to the 5DIII and look at the ISO Sensitivity plot, you can see that the D800 is a little further off the nominal value at all the settings, with more separation at the higher ISOs. In other words, the D800 lies to you a little more than the 5DIII (but the 5DIII is still lying). For example, when you set both cameras to ISO 6400, the D800 is actually shooting at ISO 4211 (it's lying by 2/3-stop), whereas the 5DIII is actually shooting at ISO 5179 (it's lying by only 1/3-stop). Translation - artificial advantage for the D800 because it's shooting at a lower actual ISO than the 5DIII for a given setting.

Bottom line is three fantastic cameras, just take your pick and be happy!

This is the take home message!

Oh, I fully agree. And as you pointed out, it further drives home the idea that you have to pay attention to what was actually posted.

 

[bp]

Nikon, going really fast.

"Nikon, going real fast"?

The DxO just tests the sensor. Sony makes the D800 sensor, not Nikon. Shouldn't all this fanboi love be giving Sony the credit? All Nikon did was stick it in a camera "system", that by all other accounts is far inferior.

So... Way to go Sony! Too bad Nikon f#@%ed it up by putting it in that body, but very nice job on the sensor.

 

[Fishnose]

1) ISO sesitivity - no meaningful differences except the total range (5D3 beats D800 and D4 beats 5D3)

Just to call this one out, DxOMark's ISO Sensitivity measurement isn't measuring what most people think it's measuring. It has no direct bearing on ISO noise performance (but an indirect one, see below). The name is perhaps misleading, maybe better to call it 'ISO Accuracy' or 'ISO Fidelity'. What this test measures is the real ISO value (benchmarked against the actual International Organization for Standardization's criteria) vs. the ISO setting on the camera, or to put it another way, it measures how much the camera lies to you when you pick a given ISO setting.

In the plot you can see that for ISO 50 and ISO 100, the dots for the 5DIII, D800, and D4 are all stacked on top of each other, and they're all at ISO 75 for both settings. What that means is all three cameras are lying to you in exactly the same way - whether you set ISO 50 or ISO 100 for your shot, the exposure is actually at around ISO 75 and then pushed or pulled by the camera as needed, although the ISO value you selected is what's actually recorded in the metadata. This lying is not new or unique - both Canon and Nikon do it routinely for fast lenses, where the incident angle of the light exceeds the refracting capability of the microlenses and exposure is 'secretly' boosted to compensate (i.e. about 1/2 stop of the light coming in at f/1.2-1.4 is not detected by a digital sensor, so the camera boosts the ISO half a stop - meaning half a stop more noise - without telling you).

Perhaps of a bit more significance is the way this plays out - if you compare just the D800 to the 5DIII and look at the ISO Sensitivity plot, you can see that the D800 is a little further off the nominal value at all the settings, with more separation at the higher ISOs. In other words, the D800 lies to you a little more than the 5DIII (but the 5DIII is still lying). For example, when you set both cameras to ISO 6400, the D800 is actually shooting at ISO 4211 (it's lying by 2/3-stop), whereas the 5DIII is actually shooting at ISO 5179 (it's lying by only 1/3-stop). Translation - artificial advantage for the D800 because it's shooting at a lower actual ISO than the 5DIII for a given setting.

No Sir, this is incorrect. Your technical explanations so far have been stellar, but here you're wrong. The 'ISO accuracy' you menton is a different measurement altogether.

Quote from DxOMark concerning what their 'Sports (Low-Light ISO)' measures:

"Sports & action photography: Low-Light ISO

....Photojournalists and action photographers often struggle with low available light and high motion. Achieving usable image quality is often difficult when pushing ISO.

When shooting a moving scene such as a sports event, action photographers’ primary objective is to freeze the motion, giving priority to short exposure time. To compensate for the lack of exposure, they have to increase the ISO setting, which means the SNR will decrease. How far can they go while keeping decent quality? Our low-light ISO metric will tell them.

The SNR indicates how much noise is present in an image compared to the actual information (signal). The higher the SNR value, the better the image looks, because details aren't drowned by noise. SNR strength is given in dB, which is a logarithmic scale: an increase of 6 dB corresponds to doubling the SNR, which equates to half the noise for the same signal.

An SNR value of 30dB means excellent image quality. Thus low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits."

 

[psolberg]

Nikon, going really fast.

"Nikon, going real fast"?

The DxO just tests the sensor. Sony makes the D800 sensor, not Nikon. Shouldn't all this fanboi love be giving Sony the credit? All Nikon did was stick it in a camera "system", that by all other accounts is far inferior.

So... Way to go Sony! Too bad Nikon f#@%ed it up by putting it in that body, but very nice job on the sensor.

apple designs the iphone, foxconn makes it. Nikon designed the sensor, sony makes it.

 

[kozakm]

apple designs the iphone, foxconn makes it. Nikon designed the sensor, sony makes it.

To be correct, Nikon designed some supporting circuitry around the sensor, not the sensor itself...

 

[skitron]

So I wonder where all the D4 owners are screaming about how D800 kills it in color sensitivity below ISO ~360? Tt's a total embarassment I say!!!

 

[Fishnose]

Nikon, going really fast.

"Nikon, going real fast"?

The DxO just tests the sensor. Sony makes the D800 sensor, not Nikon. Shouldn't all this fanboi love be giving Sony the credit? All Nikon did was stick it in a camera "system", that by all other accounts is far inferior.

So... Way to go Sony! Too bad Nikon f#@%ed it up by putting it in that body, but very nice job on the sensor.

Boringggg.....

I'm talking about Nikon working hard to trump their competitors by making smart moves, and Canon doing other things because they're too complacent and comfortable. If Canon decides to make their own sensors and loses market share because of it, that's known as a marketing mistake. Simple.

Case in point: Nikon D3200. Higher resolution than any Canon ever made and it's their CHEAPEST model in the new line. The image quality is apparently darn good. Yes, I now it's DX, but it's still remarkable - and a brilliant move.

If Canon wants to avoid a disastrous loss in market share they need to DO something, not sit on their behinds.

I've worked with R&D and I know exactly how this works, how companies do well and then relax. And lose.

 

[psolberg]

apple designs the iphone, foxconn makes it. Nikon designed the sensor, sony makes it.

To be correct, Nikon designed some supporting circuitry around the sensor, not the sensor itself...

source?

 

[hungp]

From DxOMark: "low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits"
I reviewed the chart and found though SNR chart are almost the same, 5D iii did not pass the terms "while".
The fact is who really care of color depth and dynamic range in low light real world? how many color and how much light in the dark?

I am not sure why DxO so bias like this? or is this just a trick from Nikon to break the benchmark?

 

[EYEONE]

1) ISO sesitivity - no meaningful differences except the total range (5D3 beats D800 and D4 beats 5D3)

Just to call this one out, DxOMark's ISO Sensitivity measurement isn't measuring what most people think it's measuring. It has no direct bearing on ISO noise performance (but an indirect one, see below). The name is perhaps misleading, maybe better to call it 'ISO Accuracy' or 'ISO Fidelity'. What this test measures is the real ISO value (benchmarked against the actual International Organization for Standardization's criteria) vs. the ISO setting on the camera, or to put it another way, it measures how much the camera lies to you when you pick a given ISO setting.

In the plot you can see that for ISO 50 and ISO 100, the dots for the 5DIII, D800, and D4 are all stacked on top of each other, and they're all at ISO 75 for both settings. What that means is all three cameras are lying to you in exactly the same way - whether you set ISO 50 or ISO 100 for your shot, the exposure is actually at around ISO 75 and then pushed or pulled by the camera as needed, although the ISO value you selected is what's actually recorded in the metadata. This lying is not new or unique - both Canon and Nikon do it routinely for fast lenses, where the incident angle of the light exceeds the refracting capability of the microlenses and exposure is 'secretly' boosted to compensate (i.e. about 1/2 stop of the light coming in at f/1.2-1.4 is not detected by a digital sensor, so the camera boosts the ISO half a stop - meaning half a stop more noise - without telling you).

Perhaps of a bit more significance is the way this plays out - if you compare just the D800 to the 5DIII and look at the ISO Sensitivity plot, you can see that the D800 is a little further off the nominal value at all the settings, with more separation at the higher ISOs. In other words, the D800 lies to you a little more than the 5DIII (but the 5DIII is still lying). For example, when you set both cameras to ISO 6400, the D800 is actually shooting at ISO 4211 (it's lying by 2/3-stop), whereas the 5DIII is actually shooting at ISO 5179 (it's lying by only 1/3-stop). Translation - artificial advantage for the D800 because it's shooting at a lower actual ISO than the 5DIII for a given setting.

No Sir, this is incorrect. Your technical explanations so far have been stellar, but here you're wrong. The 'ISO accuracy' you menton is a different measurement altogether.

Quote from DxOMark concerning what their 'Sports (Low-Light ISO)' measures:

"Sports & action photography: Low-Light ISO

....Photojournalists and action photographers often struggle with low available light and high motion. Achieving usable image quality is often difficult when pushing ISO.

When shooting a moving scene such as a sports event, action photographers’ primary objective is to freeze the motion, giving priority to short exposure time. To compensate for the lack of exposure, they have to increase the ISO setting, which means the SNR will decrease. How far can they go while keeping decent quality? Our low-light ISO metric will tell them.

The SNR indicates how much noise is present in an image compared to the actual information (signal). The higher the SNR value, the better the image looks, because details aren't drowned by noise. SNR strength is given in dB, which is a logarithmic scale: an increase of 6 dB corresponds to doubling the SNR, which equates to half the noise for the same signal.

An SNR value of 30dB means excellent image quality. Thus low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits."

You quoted the description for the SNR%18 result which is not what Neuro was talking about. He was explaining the ISO Sensitivity portion.

 

[hungp]

From DxOMark: "low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits"
I reviewed the chart and found though SNR chart are almost the same, 5D iii did not pass the terms "while".
The fact is who really care of color depth and dynamic range in low light real world? how many color and how much light in the dark?

I am not sure why DxO so bias like this? or is this just a trick from Nikon to break the benchmark?

I think DxOMark consider the benchmarked ISO, not the "lied" ISO.

 

[Renato]

If you are looking for good files to compare then go here:

http://www.imaging-resource.com/PRODS/nikon-d800/nikon-d800A7.HTM
and
http://www.imaging-resource.com/PRODS/canon-5d-mkiii/canon-5d-mkiiiA7.HTM

I have found this site a great source for a FAIR comparison.
My personal choice goes for 5D Mark3, I have one on order.

 

[Fishnose]

You quoted the description for the SNR%18 result which is not what Neuro was talking about. He was explaining the ISO Sensitivity portion.

The text I quoted from DxO is their own description of how they test for "Sports (Low-Light ISO)". Which is the ISO value given in the sensor scores (2293 ISO in the case of the 5DMkIII).
Read their tech texts.

 

[hungp]

You quoted the description for the SNR%18 result which is not what Neuro was talking about. He was explaining the ISO Sensitivity portion.

The text I quoted from DxO is their own description of how they test for "Sports (Low-Light ISO)". Which is the ISO value given in the sensor scores (2293 ISO in the case of the 5DMkIII).
Read their tech texts.

Now I can sure Nikon sensor "better" in Sport-Low light because of it has a better Color depth (17 vs 18) which my eyes can not see any differences, especially in the low light situation, my 5d iii is great and I just don't care DxO any more.