5DS scores at DXO **now posted**

[bdunbar79]

Wow so much hate for DXO Mark.

I have run my own tests Nikon vs Canon as myself and my family own both types of gear, and I can replicate every result that DXO produce.

I even managed to replicate an interesting flaw in their testing, one lens showed better sharpness at f16 than all others, I replicated this and discovered that this lens did not actually stop down to F16 but was still at F11 giving the erroneous result.

I have found them to be extremely reliable, and these new figures they have posted pretty much match exactly what everyone else is saying.

For example, they say that the low light score, ie the ISO performance is behind that of the D810, but very similar to that of a 5DMKIII

DP Review example pics show exactly this.

http://www.dpreview.com/previews/canon-eos-5ds-sr/5

They say the dynamic range is behind that of the D810, but improved from previous Canon cameras.
DP Review examples show the same thing.

http://www.dpreview.com/previews/canon-eos-5ds-sr/7

What all these results show is that unlike Sony and Nikon, who have managed the feat of giving you more pixels and better iso and better DR all in the same full frame package, Canon have simply scaled up the 7D MK II's crop sensor with all its compromises. It is not a great leap forward in technology, and the upcoming Sony 42Mpix backlit sensor is probably going to destroy it.

An interesting point, I tested Nikon D810 with 24-70 vs Canon 5DMK II with 24-105 lens, I found that the Nikon at 70mm resolved the same as the Canon at 105mm. As a *system* this is what matters when producing a file. And that is what DXO Mark scores showed this. They also showed that the Nikon 24-70 would suffer CA in the corners, and that is what I found.

So I expect that now there is a system that has 50mpix and some great lenses, you will see Canon lenses up top with some of the best performing sharpness scores, and I think we will see the big white telephotos finally showing just how sharp they are on a sensor that can resolve all they can give.

Then if you have so much experience with these different systems how do you explain the DxO scoring of APS sensors against FF ?

On many of the Sony or now Toshiba crop sensors DxO has them equal to, or even higher than, FF sensors. Yet if you use a 'higher rated' crop sensor alongside an 'inferior' FF sensor the FF is significantly better in terms of colour definition, tonal graduation, overall definition, practical resolution......the list goes on.

Likewise the extra DR of the Exmor sensor (and I guess Toshiba from what I have seen of the D4 data) is of little practical value outside of artificially raising shadow data, so the 'scores' are biased to this end result - which is of little use to the majority.

If you look at those scores, lets compare the 5DMK III to the D7100 for example.

http://www.dxomark.com/Cameras/Compare/Side-by-side/Canon-EOS-5DS-R-versus-Canon-EOS-5D-Mark-III-versus-Nikon-D7100___1009_795_865

You will find the scores again match up with reality, the Nikon has very very slightly better colour depth, much better DR and much worse ISO performance.

And if you look at DP Reviews published Raw files and try to push the shadows for the 5d Mk3 you will see it has a horrible performance, with lots of noise, but worse is the banding which is almost impossible to automatically remove.

And if you look at the high iso studio shots for the D7200, you find it has poor ISO performance over iso 800 and the 5DMK III twice as good - which is exactly what the DXO numbers show.

And you talk about overall resolution and practical definition, well, again if you check DXO's number you find they accurately represent this - and that the same sigma 50mm lens is sharper on the Canon Full frame cameras than it is on the APS-C Nikon camera.

http://www.dxomark.com/Lenses/Compare/Side-by-side/50mm-F1.4-EX-DG-HSM-Nikon-on-Nikon-D7100-versus-50mm-F1.4-EX-DG-HSM-Canon-on-Canon-EOS-5D-Mark-III__202_865_201_795

Again we can see with our own eyes that DXO's numbers are accurate, and in fact there are many measurements where the the Full frame camera beats the APS-C camera and that is accurately represented.

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure, but the reality is Canon are using sensor technology 10 years out of date and desperately needs to build a new chip fab plant, and this is the price they are paying for this lack of investment.

And regarding DR being useless, it is incredibly useful in the real world ! I have taken photos of birds in flight where I have been able to recover the details of their eyes and the colour of the irises, bringing the photo to life. I no longer need a fill flash to shoot in direct sunlight. I can get beautiful backlight shots and bring the eyes and the face back into balance without banding or noise or loss of skintones. I can shoot weddings in a dark church and have the couple and the sunlit stained glass window all exposed perfectly.

Looks like you missed your bus again. Nobody is talking about their measurements per se, but the SCORES. The scores are what have little value, if any. I also hate SOME of the measurements. But those at least I can read and draw the correct conclusions.

 

[krisbell]

Now, if your photographic technique is solely based on severely underexposing at ISO 100, then pushing 3+ stops in post to achieve washed out ( "recovered" ) images - Nikon/Sony is for you. No question about it.

For everything else, Canon is the choice.

Urgghh what a terrible load of nonsense. I have to always Expose to the Right when shooting with a Canon to get round the poor noise performance characteristics of the 5DIII (one of Canon's best models). With a Sony, I can choose a normal exposure, or if the scene has a massive dynamic range, I have the ability to under expose slightly, safe in the knowledge that if required I can lift shadows by several stops without running into the sort of problems that a Canon does. With Sony I have the flexibility of choice, with Canon I dont - its as simple as that, and anyone that says this isnt one advantage of a Sony sensor (albeit one whose importance will vary greatly between individuals) is deluding themselves.

I push almost every single image I produce by 3+/- stops in post in some portion/part of the image, and I have not had anyone comment about how washed out/recovered the images are.

 

[Sporgon]

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure

That composite score is by and large what people like to poke fun at. And really, that's what it is: poking fun; nobody real is genuinely offended by DXO Mark, and even those of us who find their scoring wonky/inconsistent tend to appreciate the underlying data.

Looking at what bobhope stated earlier in the full post, those measurements do not, in my experience, do not come through when making the practical application. So in other words in things like 'colour depth' it doesn't matter one jot if the crop is fractionally ahead; in practice after enlargement and data manipulation etc the FF is way out in front.

However the sentence that has been quoted here is what really drives me nuts over the "two stops extra" measurements.

All that two stops in buried in the very extreme base of the sensor; it is not linear across the data. As we all know, exposure is not linear, one stop more is twice the light and so forth. So when bobhope states that it is "two stops more so four times in fact" he would only be right if the extra DR was at the highlight end of the sensor response, and it most certainly is not. In fact his argument is the complete antithesis of what is really happening: because that extra "two stops" is right at the low light base line, it is really more akin to saying it has an extra three quarter stop overall, and this is how I find it works in the field.

How often do we hear people on this forum saying they've done a two stop lift in post and image if they had that amount extra in reserve "aka the 'two stops extra' on the Exmor sensor " ? It doesn't work like that due to the point I made above.

So I should add it is for this reason that I believe the the biase on DR that goes towards the final 'score' is very misleading to the buying public and anyone else who looks to DxO for sensor performance.

 

[surapon]

Dear Teachers and friends.
Now, I am very confuse---I Pre-Order Canon 5DSR, From My local Camera shop, for 2 month, Still not come yet---But He has 8 , 5DS in stock.
Now From this post, DXO said that, for Overall score= 5DS = 87, 5DSR= 86----WHAT WILL I DO ?
Should I forget about Canon 5DSR, and go to get 5DS to day ?.
Please help me answer this question. Because Next Month, I will have a beautiful Vacation at Yellow Stone Park and around there for 7 Days, And I want to use this Camera.
Thank you, Sir/ Madame.
Surapon

 

[ahsanford]

Finally. DXO Fever has set in.

I must admit that I am disappointed in how slowly it set in yesterday. But I went to bed, woke up, and I see the chaos bomb has well and truly gone off.

- A

 

[Don Haines]

Finally. DXO Fever has set in.

I must admit that I am disappointed in how slowly it set in yesterday. But I went to bed, woke up, and I see the chaos bomb has well and truly gone off.

- A

I have a bowl of popcorn, a fully charged iPad, a laser pointer, and a kitten.... I'm set for entertainment and waiting for someone to post squirrel pictures with 15 stops of DR.....

 

[retroreflection]

Dear Friend Surapon

DXO scores have nothing to do with it.
You have decided that a 5DSR will be the optimal next step beyond your current bodies. You also have decided that the 5DS is suboptimal by a tiny fraction. We know that you could enjoy the trip with your current gear, but you have the cash, so why not make the trip better?
It is an easy leap to say 5DS for the trip is much better than no 5DSR for the trip.

You should have a week prior to departure to get to know your new friend. That date is the drop dead date for the decision. Have a frank discussion with the shop,"If you can't deliver the R by "X", then I will take one of these in stock. Is that deadline acceptable for you, or should it be even earlier?"

If you get the S, you can just keep it and be happy with a great camera, or many on this forum would be happy to buy it at a minor discount when your R does come in.

 

[sanj]

Dear Teachers and friends.
Now, I am very confuse---I Pre-Order Canon 5DSR, From My local Camera shop, for 2 month, Still not come yet---But He has 8 , 5DS in stock.
Now From this post, DXO said that, for Overall score= 5DS = 87, 5DSR= 86----WHAT WILL I DO ?
Should I forget about Canon 5DSR, and go to get 5DS to day ?.
Please help me answer this question. Because Next Month, I will have a beautiful Vacation at Yellow Stone Park and around there for 7 Days, And I want to use this Camera.
Thank you, Sir/ Madame.
Surapon

I don't have either, but if were you I would wait for the R. This is based on what I have read on internet. Have a great time at Yellow Stone! Wow. I just returned from NY, will post few photos for all to see.

 

[ahsanford]

Finally. DXO Fever has set in.

I must admit that I am disappointed in how slowly it set in yesterday. But I went to bed, woke up, and I see the chaos bomb has well and truly gone off.

- A

I have a bowl of popcorn, a fully charged iPad, a laser pointer, and a kitten.... I'm set for entertainment and waiting for someone to post squirrel pictures with 15 stops of DR.....

Right?

I will spend my morning hooking up my iPhone + DXO One combo onto my Arca Cube, which I will then weld on to a 5000 lb ingot of steel -- because who uses tripods? Then I will take a higher quality shot of a forest than my 5D3 can -- just as soon as I erect a wind baffle as large as the Cliffs of Dover to stop those pesky leaves from moving during my 4 exposures of magic.

And did you hear, all of our lenses are now ranked 250 spots higher because we now have more pixels to rate them with! My venerable 50 f/1.4 USM now outperforms the Nikon-mount Zeiss Otus lenses. I always knew those things were overrated.

- A

 

[PureClassA]

Surapon,

I initially pre-ordered the 5DS then changed a week later to the R. I'm really happy I did. I haven't had any problems with moire or nasty edges. I was concerned with getting a lot of false detail like I have seen in many Exmor images (comparing lack of LPF), but the 5DSR looks really smooth there. I'm glad I got it because I think I would have kept wondering if I'd stuck with the 5DS. Figured if I got the R and had problems I could always sell it and get the S. Stick with what you got and don't read DxO scores. They don't take pictures. You do.

Dear Teachers and friends.
Now, I am very confuse---I Pre-Order Canon 5DSR, From My local Camera shop, for 2 month, Still not come yet---But He has 8 , 5DS in stock.
Now From this post, DXO said that, for Overall score= 5DS = 87, 5DSR= 86----WHAT WILL I DO ?
Should I forget about Canon 5DSR, and go to get 5DS to day ?.
Please help me answer this question. Because Next Month, I will have a beautiful Vacation at Yellow Stone Park and around there for 7 Days, And I want to use this Camera.
Thank you, Sir/ Madame.
Surapon

 

[BobHope]

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure

That composite score is by and large what people like to poke fun at. And really, that's what it is: poking fun; nobody real is genuinely offended by DXO Mark, and even those of us who find their scoring wonky/inconsistent tend to appreciate the underlying data.

Looking at what bobhope stated earlier in the full post, those measurements do not, in my experience, do not come through when making the practical application. So in other words in things like 'colour depth' it doesn't matter one jot if the crop is fractionally ahead; in practice after enlargement and data manipulation etc the FF is way out in front.

However the sentence that has been quoted here is what really drives me nuts over the "two stops extra" measurements.

All that two stops in buried in the very extreme base of the sensor; it is not linear across the data. As we all know, exposure is not linear, one stop more is twice the light and so forth. So when bobhope states that it is "two stops more so four times in fact" he would only be right if the extra DR was at the highlight end of the sensor response, and it most certainly is not. In fact his argument is the complete antithesis of what is really happening: because that extra "two stops" is right at the low light base line, it is really more akin to saying it has an extra three quarter stop overall, and this is how I find it works in the field.

How often do we hear people on this forum saying they've done a two stop lift in post and image if they had that amount extra in reserve "aka the 'two stops extra' on the Exmor sensor " ? It doesn't work like that due to the point I made above.

So I should add it is for this reason that I believe the the biase on DR that goes towards the final 'score' is very misleading to the buying public and anyone else who looks to DxO for sensor performance.

Sorry, you are wrong, you *can* boost exposure by an extra 2 stops or 4 times in post across the whole image.

This is how the dynamic range is calculated by DXO mark.

http://www.dxomark.com/Reviews/Canon-500D-T1i-vs.-Nikon-D5000/Dynamic-range-and-noise-source

It's the readout noise of the Canon sensors that is causing the problem. Although this article was written in 2009, it is worth reading as Canon has not really progressed since then, this is due to the chip architecture where all the ADC is performed off the chip leading to greater noise.

Back in 2009, the Canon and Nikon scores were very close, but Canon have stood still and hence their overall score has lagged. People seem to think its some sort of DXO conspiracy, but I don't think that sentiment adds up, how could they have predicted canon making almost no progress all this time ?

Some examples of DR:

Here is an image I took at iso 200 when I was testing my better beamer flash and totally screwed the exposure. Obviously not the right way to photograph but it shows you how far you can push modern sensors.

DSC_8236_Original by Piston Heads, on Flickr

Here it is pushed 4 stops in post. No noise reduction has been applied.

DSC_8236 by Piston Heads, on Flickr

Here is a real world example where I exposed correctly and still needed the Dynamic range. I was able to use the dynamic range so that I could recover both the sky and the iris colour of a bird in flight. The eyes are pushed 4 stops in post and appear totally black in the original image.

Hobby_Eye_Level_Flight Thursley Common by Nature Ist, on Flickr

Sorry, no squirrels.

 

[3kramd5]

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure

That composite score is by and large what people like to poke fun at. And really, that's what it is: poking fun; nobody real is genuinely offended by DXO Mark, and even those of us who find their scoring wonky/inconsistent tend to appreciate the underlying data.

Looking at what bobhope stated earlier in the full post, those measurements do not, in my experience, do not come through when making the practical application

Sure, real world application and laboratory measurement may not meaningfully correlate. For example, DXO suggests that the 36MP exmor has "two-thirds of a stop better image quality" than the 5DS (meaning they're boiling down "image quality" to something measurable in stops). One might conclude then that if two images are set side by side of the same scene taken exposed identically, the one recoded with the exmor will look nearly twice as good as the one taken with the 51MP canon. Clearly that's not the case.

 

[dadgummit]

DXO scores are usually rubbish but always remember this quote as it is always correct:

The best camera in the world is the one in your hands when you need to take the shot.

Dear Teachers and friends.
Now, I am very confuse---I Pre-Order Canon 5DSR, From My local Camera shop, for 2 month, Still not come yet---But He has 8 , 5DS in stock.
Now From this post, DXO said that, for Overall score= 5DS = 87, 5DSR= 86----WHAT WILL I DO ?
Should I forget about Canon 5DSR, and go to get 5DS to day ?.
Please help me answer this question. Because Next Month, I will have a beautiful Vacation at Yellow Stone Park and around there for 7 Days, And I want to use this Camera.
Thank you, Sir/ Madame.
Surapon

 

[3kramd5]

Sorry, you are wrong, you *can* boost exposure by an extra 2 stops or 4 times in post across the whole image.

Whether you can boost exposure across the whole image depends on where the tones lie; you can't quadruple the exposure of saturated pixels. With low read noise, you can meaningfully (i.e. without noise becoming subjectively destructive) boost the shadows from sony sensors more than from canon sensors.

 

[3kramd5]

...
But even some of the data makes me go "huh."

For example, for testing dynamic range, they use a pretty nifty rig with controlled lighting and a series of ND filters. From their description: "We use filters having different light absorption levels ranging from 0% to 99.99% in order to test across a dynamic range of 4 density steps (= 13.3 f-stops — a dynamic range much greater than today’s digital cameras)."

But then their results show several cameras with DR significantly higher than their testing methodology can possibly measure (e.g. 14.8 f-stops)

Want to bet that they wrote that blurb about testing a higher dynamic range than "today's digital cameras" 10 years ago (or whenever) and haven't updated since the D8x0, etc?

It's possible, unfortunately that page isn't dated, and I can't find any updates to their test equipment. It's possible they don't even test it optically anymore.

 

[Sporgon]

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure

That composite score is by and large what people like to poke fun at. And really, that's what it is: poking fun; nobody real is genuinely offended by DXO Mark, and even those of us who find their scoring wonky/inconsistent tend to appreciate the underlying data.

Looking at what bobhope stated earlier in the full post, those measurements do not, in my experience, do not come through when making the practical application

Sure, real world application and laboratory measurement may not meaningfully correlate. For example, DXO suggests that the 36MP exmor has "two-thirds of a stop better image quality" than the 5DS (meaning they're boiling down "image quality" to something measurable in stops). One might conclude then that if two images are set side by side of the same scene taken exposed identically, the one recoded with the exmor will look nearly twice as good as the one taken with the 51MP canon. Clearly that's not the case.

This is the crux of the matter. ^^

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure

That composite score is by and large what people like to poke fun at. And really, that's what it is: poking fun; nobody real is genuinely offended by DXO Mark, and even those of us who find their scoring wonky/inconsistent tend to appreciate the underlying data.

Looking at what bobhope stated earlier in the full post, those measurements do not, in my experience, do not come through when making the practical application. So in other words in things like 'colour depth' it doesn't matter one jot if the crop is fractionally ahead; in practice after enlargement and data manipulation etc the FF is way out in front.

However the sentence that has been quoted here is what really drives me nuts over the "two stops extra" measurements.

All that two stops in buried in the very extreme base of the sensor; it is not linear across the data. As we all know, exposure is not linear, one stop more is twice the light and so forth. So when bobhope states that it is "two stops more so four times in fact" he would only be right if the extra DR was at the highlight end of the sensor response, and it most certainly is not. In fact his argument is the complete antithesis of what is really happening: because that extra "two stops" is right at the low light base line, it is really more akin to saying it has an extra three quarter stop overall, and this is how I find it works in the field.

How often do we hear people on this forum saying they've done a two stop lift in post and image if they had that amount extra in reserve "aka the 'two stops extra' on the Exmor sensor " ? It doesn't work like that due to the point I made above.

So I should add it is for this reason that I believe the the biase on DR that goes towards the final 'score' is very misleading to the buying public and anyone else who looks to DxO for sensor performance.

Sorry, you are wrong, you *can* boost exposure by an extra 2 stops or 4 times in post across the whole image.

This is how the dynamic range is calculated by DXO mark.

http://www.dxomark.com/Reviews/Canon-500D-T1i-vs.-Nikon-D5000/Dynamic-range-and-noise-source

It's the readout noise of the Canon sensors that is causing the problem. Although this article was written in 2009, it is worth reading as Canon has not really progressed since then, this is due to the chip architecture where all the ADC is performed off the chip leading to greater noise.

Back in 2009, the Canon and Nikon scores were very close, but Canon have stood still and hence their overall score has lagged. People seem to think its some sort of DXO conspiracy, but I don't think that sentiment adds up, how could they have predicted canon making almost no progress all this time ?

Some examples of DR:

Here is an image I took at iso 200 when I was testing my better beamer flash and totally screwed the exposure. Obviously not the right way to photograph but it shows you how far you can push modern sensors.

DSC_8236_Original by Piston Heads, on Flickr

Here it is pushed 4 stops in post. No noise reduction has been applied.

DSC_8236 by Piston Heads, on Flickr

Here is a real world example where I exposed correctly and still needed the Dynamic range. I was able to use the dynamic range so that I could recover both the sky and the iris colour of a bird in flight. The eyes are pushed 4 stops in post and appear totally black in the original image.

Hobby_Eye_Level_Flight Thursley Common by Nature Ist, on Flickr

Sorry, no squirrels.

I would never dispute that in this particular exposure situation the exmor isn't far ahead. It is. Simple.

However this does not mean that in even a moderately exposed scene the 'IQ' of the exmor / toshiba etc is better than Canon. 3kramd5 summed it up perfectly.

As far as I'm aware dxo don't advertise 'landscape' score as 'boy did I really f**k up' score.

 

[sanj]

...
Remember folks, DXO rates the D3300 better than the 1DX. That's all you need to know.
...

And is there any reason why the D3300 can't outscore the 1DX in these tests?

If I build the a camera with the perfect sensor with 100% QE, no noise at ISO 100-6400 and put that in a body like the SL1's and sell it for $600, why shouldn't it score better than a 1DX? Does the 1DX's sensor have to score more just because it costs more? Or because it is bigger?

It's really strange that people have an inbuilt assumption that bigger and more expensive necessarily means better when it comes to DxO scores. It simply isn't the case. Not for lenses, not for camera sensors.

I agree with your logic. Really do! But I prefer the speed of the 1dx to get me the fast moving shots I like to photograph. Getting the shot is that matters to me with modern sensors, the difference is not much comparing 'equal' cameras. I have never logged on to DxO website, but would want Canon to lead in all tests everywhere.

 

[scyrene]

Huh? Every thread in which they're mentioned? Read back through this one even. I just want to hear a cogent defence of their methodology, rather than a 'critics are all fanboys' ad hominem.

You want a defense of the methodology? How about this.

In the lens testing, the most heavily weighted factor is the T-stop value. This is done because everyone knows that a faster lens is better than a slower lens and this helps the lens rating metric to properly reflect that.

Since fast lenses are obviously better than slow lenses, it only makes sense that the 50F1.8 (Canon's lowest lost lens) would be rated better than the 600F4.0 II lens (Canon's most expensive lens) and fortunately, DXO numbers show this to be true.

LOL.

 

[bholliman]

In all of this, NOBODY! denies that Sony/Nikon has more DR than Canon. Likewise, though depending on the type of photography and conditions you shoot in, the utility of having more DR ranges from "it will have little effect" to "gotta have it" there isn't anyone who would complain if the DR was increased.

Nikon having the best DR does not make it the best camera. Likewise, Canon having the best AF does not make it the best camera. You can not focus on one aspect and use it to represent the entirety. It is as failed of a concept as the earlier lens example where we fixated on aperture and came to the conclusion that the 50F1.8 is superior to the 600F4IS II. There are so many variable to consider that the statement is meaningless. You can't even say that one is better than the other for birding.... is the bird 200 feet away or is it 2 feet away?

The problem is boiling a camera (or a lens) down to a single metric and not revealing how that metric was obtained.

Since photography is a collection of diverse goals, needs, and resources.... having a single metric means that the score will be biased. If you calculate the metric based on the needs of a landscape photographer, it will be biased against studio photographers. If you calculate based on the needs of studio photographers, it will be biased against wedding photographers..... and so on.... and so on..... and so on.... You can't win!

You people can argue the details until the cows come home and there will be no resolution. The problem is the concept. The details are meaningless under a faulty concept.

+1 Well stated Don

 

[scyrene]

The "headline" DXO figure appears weighted towards DR, but the reality is 2 stops of DR is 4 times better, so it does skew the headline figure

That composite score is by and large what people like to poke fun at. And really, that's what it is: poking fun; nobody real is genuinely offended by DXO Mark, and even those of us who find their scoring wonky/inconsistent tend to appreciate the underlying data.

Looking at what bobhope stated earlier in the full post, those measurements do not, in my experience, do not come through when making the practical application. So in other words in things like 'colour depth' it doesn't matter one jot if the crop is fractionally ahead; in practice after enlargement and data manipulation etc the FF is way out in front.

However the sentence that has been quoted here is what really drives me nuts over the "two stops extra" measurements.

All that two stops in buried in the very extreme base of the sensor; it is not linear across the data. As we all know, exposure is not linear, one stop more is twice the light and so forth. So when bobhope states that it is "two stops more so four times in fact" he would only be right if the extra DR was at the highlight end of the sensor response, and it most certainly is not. In fact his argument is the complete antithesis of what is really happening: because that extra "two stops" is right at the low light base line, it is really more akin to saying it has an extra three quarter stop overall, and this is how I find it works in the field.

How often do we hear people on this forum saying they've done a two stop lift in post and image if they had that amount extra in reserve "aka the 'two stops extra' on the Exmor sensor " ? It doesn't work like that due to the point I made above.

So I should add it is for this reason that I believe the the biase on DR that goes towards the final 'score' is very misleading to the buying public and anyone else who looks to DxO for sensor performance.

Sorry, you are wrong, you *can* boost exposure by an extra 2 stops or 4 times in post across the whole image.

This is how the dynamic range is calculated by DXO mark.

http://www.dxomark.com/Reviews/Canon-500D-T1i-vs.-Nikon-D5000/Dynamic-range-and-noise-source

It's the readout noise of the Canon sensors that is causing the problem. Although this article was written in 2009, it is worth reading as Canon has not really progressed since then, this is due to the chip architecture where all the ADC is performed off the chip leading to greater noise.

Back in 2009, the Canon and Nikon scores were very close, but Canon have stood still and hence their overall score has lagged. People seem to think its some sort of DXO conspiracy, but I don't think that sentiment adds up, how could they have predicted canon making almost no progress all this time ?

Some examples of DR:

Here is an image I took at iso 200 when I was testing my better beamer flash and totally screwed the exposure. Obviously not the right way to photograph but it shows you how far you can push modern sensors.

DSC_8236_Original by Piston Heads, on Flickr

Here it is pushed 4 stops in post. No noise reduction has been applied.

DSC_8236 by Piston Heads, on Flickr

Here is a real world example where I exposed correctly and still needed the Dynamic range. I was able to use the dynamic range so that I could recover both the sky and the iris colour of a bird in flight. The eyes are pushed 4 stops in post and appear totally black in the original image.

Hobby_Eye_Level_Flight Thursley Common by Nature Ist, on Flickr

Sorry, no squirrels.

Nice hobby On a practical level, you can recover bird eye irises with Canon images, as they are so small and smooth you can aggressively raise the shadows and denoise on the small area of the eye, I do that sometimes. The result overall is similar.