Return of Interesting article on DxO about 5D III and D800...

Started Apr 10, 2013 | Discussions
John Sheehy
Forum ProPosts: 15,999
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to qianp2k, Apr 13, 2013

qianp2k wrote:

John Sheehy wrote:

RichRMA wrote:

If that were absolute, the Pentax Q would be the most important wildlife/astrocamera extant.

The only reason that I don't use the Q for more focal-length-limited work is because it is not easy to use it adapted to DSLR lenses, as far as subject quality is concerned, versus cropping from DSLR sensors, it trashes them.  The big pixels just can't compete, for both resolution and visible noise.

John, I know you're a huge fan of super high pixel density

Anyone who truly understands what proper sampling is, does; the only drawback is speed and storage when high pixel density is combined with large sensors, and that may vary opinions a bit.

. This scenario only applied when you unable to move closer or you don't have long enough lenses.

This scenario is very common for certain types of shooting.  "Only" disappears in that context.  Skittish birds like wild sparrows can often only be approached, even carefully, down to about 30 - 40 feet.  Do you know how small a small sparrow is on an APS-c frame at that distance with 400mm?  It would fit in a cellphone sensor, with room to spare.  Warblers at certain times of migration forage mainly high in the trees, with only a rare individual coming down to a decent level.  You need to get far from a tree to have a good perspective on a warbler foraging in the treetops in the spring.

If there was an EOS camera, fully functional like the 7D but with a much smaller sensor with much smaller pixels, I'd pay $2000 for it in an instant.

However in reality Pro(s) and many enthusiasts willing to spend more on super tele lenses and DSLRs in various crop formats so they don't need to crop severely.

The range of crop formats is way too narrow.  I would like to see formats large enough that the entire image circle of FF lenses are in the frame, so you can crop to the max of the circle.  I'd also like to be able to use my EOS lenses on smaller formats than APS-c.  A smaller sensor, unless the implementation is totally botched, noise-wise, is superior to using a TC if the need for shutter speed prevents you from getting a full base-ISO exposure.  That's the only case in which using a TC is superior, because you can saturate the full sensor, collecting more photons in a single exposure.  Most wildlife photographers use ISO 400 to 1600, but you can collect as much light with the small sensor at a lower ISO forgoing the TC, and by not cropping away total sensor light collection.

Then no way Q or any small sensor cameras can even touch Sharpness (resolution + acutance) and IQ as a whole.

You're regressing to a different paradigm.  We're talking about when one is focal-length-limited, and in that context, only resolution and noise per unit of sensor area are relevant.   The Q's sensor trashes any FF sensor in that context (and there's nothing special about the Q's sensor; it is just a run-of-the-mill 1/2.3" sensor that happens to have RAW output and can be adapted to a plethora of lenses).  DR is about the same as the D800, and high-ISO noise is only about 1/2 stop behind the best DSLRs, and better than the worst.  Resolution potential is high, and some good DSLR lenses actually are under-sampled by the Q's pixel density (like my 70-200/4L IS at 200mm and f/4).

You seem to fail to appreciate the differences between sensor per unit of sensor area; all you seem to see is pixels.  I don't want to see that there are any pixels at all, in my captures!  I want virtual analog capture; you seem to want crispy mosaics.  Think of film, as an analogy.  You can put the same emulsion in a 110 compact film camera, and an 8x10" larger format.  The 8x10" gives much better resolution with equitable lenses, and has less appearance of grain with the same display size.  With the same lens, a crop of the 8x10" the size of the 110 film frame would be exactly the same quality.   Put an inferior, grainy film in the 8x10" camera only, and it may still be superior for full image quality, but is inferior for the crop scenario - this is the analogy that most closely correlates to the Q vs the DSLR.  The Q has better "emulsion", so the inferior "emulsion" of the DSLR can be better or worse in practice, depending on how much of the DSLR frame you actually use for your output.  If you can't accept this fact, you're always going to come to false conclusions.

That's why Olympus FT DSLRs never really take off in sport fields and serious wildlife photography (except birding involved in small birds as even 800L is not long enough), no mention small-sensor Q as IQ from respective systems are not even close.

IQ is meaningless without a context.  "SQ" (subject quality) is what I worry about.  What good is is IQ if you only need a fraction of the "image".  My 6D has better IQ than my Q using the whole frame for the final output, at all ISOs.  I can't say the same for my other DSLRs.  The Q has better shadows at low ISOs, by a good margin, than my other DSLRs, including the 5D2, and it has less ugly noise at ISO 1600 than my 10D, and maybe even my 20D or 30D.  Crop those DSLRs, and things get even worse for the final result (subject displayed at a certain size).

Reply   Reply with quote   Complain
Rick Knepper
Veteran MemberPosts: 9,668Gear list
Like?
I agree completely
In reply to Mako2011, Apr 13, 2013

Mako2011 wrote:

The problem, in these types of discussions, is that folks keep mixing related terms an using them to describe inaccurately a concept

thoguh I didn't get that sense from Mr. Sheehy's post. But truth be told, I didn't understand some of it.  My problem, not his error.

But sloppy language is a huge problem in digital imaging forums and Resolution is right at the top of the list of the most mis-used terms. Resolution=print big. Naw, maybe.

-- hide signature --

Rick Knepper, photographer, non-professional, shooting for pleasure, check my profile for gear list and philosophy.

 Rick Knepper's gear list:Rick Knepper's gear list
Nikon D3X Nikon D800E Canon EOS 5D Mark III Canon EOS 6D Canon TS-E 17mm f/4L +16 more
Reply   Reply with quote   Complain
Just another Canon shooter
Senior MemberPosts: 2,071Gear list
Like?
Re: just lines
In reply to John Sheehy, Apr 13, 2013

John Sheehy wrote:

Just another Canon shooter wrote:

Why is the slanted edge test slanted then, if the angle has no bearing on resolution? What if you chose different spatial frequency and the other lens seem to win, etc.?

Of course, if the optics are properly sampled, the only difference in angular results are those imposed by the optics, and no appreciable difference is imposed by the discrete sampling structure.

I guess by properly sampled you mean well oversampled, and then I fully agree. If you just satisfy the sampling theorem, hen you have a problem. A properly sampled image can be reconstructed by a formula (given by the sampling theorem). Measuring the sensor output at spatial frequencies close to Nyquist is very different. You need an image with limiting resolution several times smaller than the pixel size to forget about the pixel structure of the image. The Nokia 808 gets closer to that, from what I have seen. And I am not even talking about color resolution. There is a reason for the slanted edge to be slanted.

 Just another Canon shooter's gear list:Just another Canon shooter's gear list
Samsung Galaxy S III
Reply   Reply with quote   Complain
John Sheehy
Forum ProPosts: 15,999
Like?
Re: John, your take on the latest Canons
In reply to Earthlight, Apr 13, 2013

Earthlight wrote:

Have you had the chance to see whether anything has changed in the latest 18 MP sensors in terms of pattern noise?

I haven't paid attention to all of them.  The 7D was improved much over the 50D I had previously, except for the vertical banding visible mostly at low ISOs, and the 60D seemed to be better in general.  I have not paid attention to the other 18MP sensors, to tell you the truth, but it seems that they are more like the 60D than the 7D.

The 6D, of course, reduced it much more, but it is not completely absent.  Pushed to ISO 1 million from 102,400, the 6D shows some fairly bold banding, and you can see a very faint 7D-like vertical corduroy pattern, but you really have to look for it, and there are occasional little dashes or short lines of banding horizontally, but I'm talking about pushing ISO 100 to 6400 or more here.

I'm far less shadow-shy at base ISOs with the 6D than with my previous Canons.  If I wanted to pull up shadows at base ISO to cover specular highlights when I was carrying my 5D2, I would pull out one of my compacts with RAW; that's how bad the 5D2 was, and the 10D, 20D, 30D, 50D were all bad for this too; the 7D a bit better, though (and part of the 7D's vertical pattern was fixed, and could potentially be removed in PP).

Reply   Reply with quote   Complain
Earthlight
Veteran MemberPosts: 3,139
Like?
Thanks! Sounds like the 6D is a nice camera then
In reply to John Sheehy, Apr 13, 2013

Six months or so ago I got the 1Ds MkIII and found it the best body by far in terms of shadow noise performance out of the cameras I have previously owned. It is good to hear that some progress has been made in this front In the latest cameras. That said, I have a 60D too and at least mine easily exhibits banding. Hopefully Canon will up its sensor game.

But should I need a second full frame body I think I would get a 6D. Judging by what you said, it should be good value.

Thanks again!

Earthlight

-- hide signature --

__________________
http://jari.pic.fi

Reply   Reply with quote   Complain
KLO82
Contributing MemberPosts: 666
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to Steen Bay, Apr 13, 2013

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:

aftab wrote:

Now, let us look at scenario B and scenario C again. In both cases A looks sharper than B. What does that mean? More resolution = more sharpness. That is resolution = sharpness in this two scenarios. And almost all scenarios are like this. So, we can actually use the terms resolution and sharpness interchangeably.

But if comparing for example the 12.7mp 5D vs. the 18mp 7D, then things get a bit more complicated, because it seems that 5D (most often, with most lenses) has the best sharpness (MTF-50?),

true.

while the 7D has the highest resolution (MTF-10 or MTF-5?).

Not necessarily. Actually with most  EF lenses 5D outresolves 7D if you frame them in the same AOV (as tested by DXOMark in either old MTF or new P-MPix unit that are two different expressing units but with the same result). Only very few lenses such as with 300L/2.8 IS II 7D/60D outresolves 5D.

-- hide signature --

But what if we could compare the MTF-10 er MTF-5 values (which we unfortunately can't anymore)? Then I think that the 7D would win, because it has quite a bit more MPs.

This is what you are talking about:

In the 2nd pic, I was supposed to highlight the MTF 20, but instead wrongly circled the MTF 30. Dxo used MTF 20 as resolution ("critical" resoltion or something like that. I cant remember the exact term).

Reply   Reply with quote   Complain
Mako2011
Mako2011 MOD
Forum ProPosts: 15,099
Like?
simple
In reply to Just another Canon shooter, Apr 13, 2013

Just another Canon shooter wrote:

Mako2011 wrote:

Just another Canon shooter wrote:

Mako2011 wrote:

Mako2011 wrote:

Image resolution is the detail an image holds. Higher resolution means more image detail.

You have to explain now what "detail" is, and what it means for an image to "hold it".

As I pointed out in this example... Number of lines. Simple stuff. You get the same count either picture.

At what angle? What contrast? Which color channel? Even if you answer all that - why would that be a measure of resolution, and not of aliasing? Why not circles instead of lines? Or points?

In scenario B, as presented here, only the line count mattered as that was the only detail present. A and B were in fact the same regards resolution. Pretty obvious that angle, contrast, and color channel had no bearing on resolution in that example.

Then you have a definition of resolution which applies to one example only?

Yes, I have only been discussing the specific scenario b in this line, as it was presented. I found it a good simple example well made for diffusion. Makes it easy to see the difference between sharpness and resolution from a common perspective

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Reply   Reply with quote   Complain
qianp2k
Senior MemberPosts: 9,527Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to KLO82, Apr 13, 2013

KLO82 wrote:

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:

aftab wrote:

Now, let us look at scenario B and scenario C again. In both cases A looks sharper than B. What does that mean? More resolution = more sharpness. That is resolution = sharpness in this two scenarios. And almost all scenarios are like this. So, we can actually use the terms resolution and sharpness interchangeably.

But if comparing for example the 12.7mp 5D vs. the 18mp 7D, then things get a bit more complicated, because it seems that 5D (most often, with most lenses) has the best sharpness (MTF-50?),

true.

while the 7D has the highest resolution (MTF-10 or MTF-5?).

Not necessarily. Actually with most  EF lenses 5D outresolves 7D if you frame them in the same AOV (as tested by DXOMark in either old MTF or new P-MPix unit that are two different expressing units but with the same result). Only very few lenses such as with 300L/2.8 IS II 7D/60D outresolves 5D.

-- hide signature --

But what if we could compare the MTF-10 er MTF-5 values (which we unfortunately can't anymore)? Then I think that the 7D would win, because it has quite a bit more MPs.

This is what you are talking about:

In the 2nd pic, I was supposed to highlight the MTF 20, but instead wrongly circled the MTF 30. Dxo used MTF 20 as resolution ("critical" resoltion or something like that. I cant remember the exact term).

Thanks for saving the snapshots of old DXOMark test in MTF unit before DXO changed to P-MPix unit.  Regardless of what unit DXOMark used, it shows the same clear result.  So P-MPix should not be an excuse for those deny/challenge DXOMark tests as from what I have seen and remembered the test comparison results are consistent with former MTF and the latter P-MPix.

With one of the best lenses 100L, we can see both systems are very close but 5D still edges out a little bit.

With a bit lesser lens 85/1.8 (still very good lens), 5D leads wider.  With further inferior lenses, then 5D will further widen the lead.  The result is clear as we have seen from tons of real world photos from respective cameras.

-- hide signature --
 qianp2k's gear list:qianp2k's gear list
Sony RX100 Canon EOS-1D Mark III Canon EOS 5D Mark II Canon EOS 5D Mark III Sony Alpha 7R +20 more
Reply   Reply with quote   Complain
schmegg
schmegg MOD
Senior MemberPosts: 4,980
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to qianp2k, Apr 13, 2013

qianp2k wrote:

The result is clear as we have seen from tons of real world photos from respective cameras.

In terms of resolution, I've not seen that. And you've been unable to show this yourself with your 5D and 60D. Though I can easily show an 18MP crop image will resolve more of the detail projected onto it than a 22MP FF.

Why do you think this is? Could it possibly be that you don't know how to read those lines properly I wonder? For instance, did you realise that both those graphs show that the crop resolves more than the FF but with less acutance? I guess not.

Once again it goes to show the folly of using lens tests to draw conclusions regarding the resolving capabilities of different sensors. All those MTF curves will tell you, when it comes to comparing different sensors, is that lower resolving sensors will give you sharper images from lenses of lesser quality than higher resolving sensors - and that's what you can see with the 85/1.8 - and why the MTF lines are much closer with the 100/2.8.

The tests are measuring acutance and it stands to reason that larger pixels will give increasingly sharper images at pixel level than smaller ones as the lens quality drops. It actually says very little about the relative resolving capabilities of the sensor and a whole lot more about the quality of the lens (surprise, surprise! Who'd have thought that a lens test would be measuring the performance of a lens! LOL!)

BTW - I can now see why DxO don't publish them any more! This is a good example of why they tried to come up with something a little easier for people to understand!

Reply   Reply with quote   Complain
Steen Bay
Veteran MemberPosts: 6,188
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to qianp2k, Apr 13, 2013

qianp2k wrote:

KLO82 wrote:

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:

aftab wrote:

Now, let us look at scenario B and scenario C again. In both cases A looks sharper than B. What does that mean? More resolution = more sharpness. That is resolution = sharpness in this two scenarios. And almost all scenarios are like this. So, we can actually use the terms resolution and sharpness interchangeably.

But if comparing for example the 12.7mp 5D vs. the 18mp 7D, then things get a bit more complicated, because it seems that 5D (most often, with most lenses) has the best sharpness (MTF-50?),

true.

while the 7D has the highest resolution (MTF-10 or MTF-5?).

Not necessarily. Actually with most  EF lenses 5D outresolves 7D if you frame them in the same AOV (as tested by DXOMark in either old MTF or new P-MPix unit that are two different expressing units but with the same result). Only very few lenses such as with 300L/2.8 IS II 7D/60D outresolves 5D.

-- hide signature --

But what if we could compare the MTF-10 er MTF-5 values (which we unfortunately can't anymore)? Then I think that the 7D would win, because it has quite a bit more MPs.

This is what you are talking about:

In the 2nd pic, I was supposed to highlight the MTF 20, but instead wrongly circled the MTF 30. Dxo used MTF 20 as resolution ("critical" resoltion or something like that. I cant remember the exact term).

Thanks for saving the snapshots of old DXOMark test in MTF unit before DXO changed to P-MPix unit.  Regardless of what unit DXOMark used, it shows the same clear result.  So P-MPix should not be an excuse for those deny/challenge DXOMark tests as from what I have seen and remembered the test comparison results are consistent with former MTF and the latter P-MPix.

Yes, thanks to KLO82 for that.

With one of the best lenses 100L, we can see both systems are very close but 5D still edges out a little bit.

With a bit lesser lens 85/1.8 (still very good lens), 5D leads wider.  With further inferior lenses, then 5D will further widen the lead.  The result is clear as we have seen from tons of real world photos from respective cameras.

-- hide signature --

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

Reply   Reply with quote   Complain
Just another Canon shooter
Senior MemberPosts: 2,071Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to Steen Bay, Apr 13, 2013

Steen Bay wrote:

qianp2k wrote:

With one of the best lenses 100L, we can see both systems are very close but 5D still edges out a little bit.

With a bit lesser lens 85/1.8 (still very good lens), 5D leads wider.  With further inferior lenses, then 5D will further widen the lead.  The result is clear as we have seen from tons of real world photos from respective cameras.

-- hide signature --

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

As I said in another thread a few weeks ago, the extinct resolution of the 7D is higher but that is at a very low contrast. In most of the contrast range, the 5D is either equal (despite having less pixels), or wins easily. I am not sure why this should be surprising to anybody. The  crop penalty is real, and the (low) mp penalty is real, too.

BTW, the real test should be at equivalent FL and equivalent f-stops with common lenses. The 5D will win by even more from MTF20 up, until you get to diffraction limited apertures. The 7D will still be able to show some measurable contrast behind the 5D Nyquist frequency.

When you go to high ISO, even to high equivalent ISOs, the advantage of the 7D will be buried in the noise. On the other hand, the 5D is noisier with equivalent ISOs.

Another observation is that DXO (now) shows the 5D to be clearly the winner in the 85/1.8 case. You can reverse engineer this to make the educated guess that their P-mpix score is weighed more towards medium contrast performance.

 Just another Canon shooter's gear list:Just another Canon shooter's gear list
Samsung Galaxy S III
Reply   Reply with quote   Complain
Dave Luttmann
Forum ProPosts: 12,327Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to Steen Bay, Apr 13, 2013

Steen Bay wrote:

qianp2k wrote:

KLO82 wrote:

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:

aftab wrote:

Now, let us look at scenario B and scenario C again. In both cases A looks sharper than B. What does that mean? More resolution = more sharpness. That is resolution = sharpness in this two scenarios. And almost all scenarios are like this. So, we can actually use the terms resolution and sharpness interchangeably.

But if comparing for example the 12.7mp 5D vs. the 18mp 7D, then things get a bit more complicated, because it seems that 5D (most often, with most lenses) has the best sharpness (MTF-50?),

true.

while the 7D has the highest resolution (MTF-10 or MTF-5?).

Not necessarily. Actually with most  EF lenses 5D outresolves 7D if you frame them in the same AOV (as tested by DXOMark in either old MTF or new P-MPix unit that are two different expressing units but with the same result). Only very few lenses such as with 300L/2.8 IS II 7D/60D outresolves 5D.

-- hide signature --

But what if we could compare the MTF-10 er MTF-5 values (which we unfortunately can't anymore)? Then I think that the 7D would win, because it has quite a bit more MPs.

This is what you are talking about:

In the 2nd pic, I was supposed to highlight the MTF 20, but instead wrongly circled the MTF 30. Dxo used MTF 20 as resolution ("critical" resoltion or something like that. I cant remember the exact term).

Thanks for saving the snapshots of old DXOMark test in MTF unit before DXO changed to P-MPix unit.  Regardless of what unit DXOMark used, it shows the same clear result.  So P-MPix should not be an excuse for those deny/challenge DXOMark tests as from what I have seen and remembered the test comparison results are consistent with former MTF and the latter P-MPix.

Yes, thanks to KLO82 for that.

With one of the best lenses 100L, we can see both systems are very close but 5D still edges out a little bit.

With a bit lesser lens 85/1.8 (still very good lens), 5D leads wider.  With further inferior lenses, then 5D will further widen the lead.  The result is clear as we have seen from tons of real world photos from respective cameras.

-- hide signature --

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

Yup.  Guess that was missed by Qianp2k because he didnt know how to read the chart.  This chart, and his own photos show the 7 D offers greater resolution.

 Dave Luttmann's gear list:Dave Luttmann's gear list
Canon PowerShot G3 Canon PowerShot SX150 IS Canon EOS D30 Canon EOS 10D Nikon D2X +15 more
Reply   Reply with quote   Complain
Mako2011
Mako2011 MOD
Forum ProPosts: 15,099
Like?
distance
In reply to Just another Canon shooter, Apr 13, 2013

Just another Canon shooter wrote:

BTW, the real test should be at equivalent FL and equivalent f-stops with common lenses.

When comparing resolution across formats you have to keep the subject distance constant. DxO apparently doesn't do that. Might be a source of some of their inconsistencies

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Reply   Reply with quote   Complain
qianp2k
Senior MemberPosts: 9,527Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to Steen Bay, Apr 13, 2013

Steen Bay wrote:

qianp2k wrote:

KLO82 wrote:

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:

aftab wrote:

Now, let us look at scenario B and scenario C again. In both cases A looks sharper than B. What does that mean? More resolution = more sharpness. That is resolution = sharpness in this two scenarios. And almost all scenarios are like this. So, we can actually use the terms resolution and sharpness interchangeably.

But if comparing for example the 12.7mp 5D vs. the 18mp 7D, then things get a bit more complicated, because it seems that 5D (most often, with most lenses) has the best sharpness (MTF-50?),

true.

while the 7D has the highest resolution (MTF-10 or MTF-5?).

Not necessarily. Actually with most  EF lenses 5D outresolves 7D if you frame them in the same AOV (as tested by DXOMark in either old MTF or new P-MPix unit that are two different expressing units but with the same result). Only very few lenses such as with 300L/2.8 IS II 7D/60D outresolves 5D.

-- hide signature --

But what if we could compare the MTF-10 er MTF-5 values (which we unfortunately can't anymore)? Then I think that the 7D would win, because it has quite a bit more MPs.

This is what you are talking about:

In the 2nd pic, I was supposed to highlight the MTF 20, but instead wrongly circled the MTF 30. Dxo used MTF 20 as resolution ("critical" resoltion or something like that. I cant remember the exact term).

Thanks for saving the snapshots of old DXOMark test in MTF unit before DXO changed to P-MPix unit.  Regardless of what unit DXOMark used, it shows the same clear result.  So P-MPix should not be an excuse for those deny/challenge DXOMark tests as from what I have seen and remembered the test comparison results are consistent with former MTF and the latter P-MPix.

Yes, thanks to KLO82 for that.

With one of the best lenses 100L, we can see both systems are very close but 5D still edges out a little bit.

With a bit lesser lens 85/1.8 (still very good lens), 5D leads wider.  With further inferior lenses, then 5D will further widen the lead.  The result is clear as we have seen from tons of real world photos from respective cameras.

-- hide signature --

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

I'd suggest everyone reads these articles to understand these concepts first and understand what that means in real world photos.

http://www.normankoren.com/Tutorials/MTF.html

http://www.imatest.com/docs/sharpness/

From above two DXOMark MTF tests, we can see 7D only extends at MTF 15% or below that virtually invisible by eyes in real world photos even at base ISO when 7D photos buried by higher noises/grains. So that 7D higher resolution (18mp vs 12.8mp from 5D) is only on paper or reflected in larger CR2 file size but unless you have an eagle eye, you just cannot see it

More meaningful MTF 50% or at least MTF 30% are truly what your eyes will enjoy   That's main reason why DXOMark changes to "perceptual sharpness" measurement as most people truly confused by MTF data, which MTF data?

I have some quotes from above two links,

Contrast levels from 100% to 2% are illustrated on the right for a variable frequency sine pattern. Contrast is moderately attenuated for MTF = 50% and severely attenuated for MTF = 10%. The 2% pattern is visible only because viewing conditions are favorable: it is surrounded by neutral gray, it is noiseless (grainless), and the display contrast for CRTs and most LCD displays is relatively high. It could easily become invisible under less favorable conditions.

The eye is relatively insensitive to detail at spatial frequencies where MTF is low: 10% or less

This measurement, also called “vanishing resolution”, corresponds to an MTF of roughly 10-20%. Because this is the spatial frequency where image information disappears— where it isn’t visible, and because it is strongly dependent on observer bias, it’s a poor indicator of image sharpness.

-- hide signature --
 qianp2k's gear list:qianp2k's gear list
Sony RX100 Canon EOS-1D Mark III Canon EOS 5D Mark II Canon EOS 5D Mark III Sony Alpha 7R +20 more
Reply   Reply with quote   Complain
Steen Bay
Veteran MemberPosts: 6,188
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to qianp2k, Apr 13, 2013

qianp2k wrote:

Steen Bay wrote:/

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

I'd suggest everyone reads these articles to understand these concepts first and understand what that means in real world photos.

http://www.normankoren.com/Tutorials/MTF.html

http://www.imatest.com/docs/sharpness/

From above two DXOMark MTF tests, we can see 7D only extends at MTF 15% or below that virtually invisible by eyes in real world photos even at base ISO when 7D photos buried by higher noises/grains. So that 7D higher resolution (18mp vs 12.8mp from 5D) is only on paper or reflected in larger CR2 file size but unless you have an eagle eye, you just cannot see it

A bit of sharpening makes it easier to see. It's clearly visible if comparing the resolution test chart shots in DPR's 5D and 7D reviews.

More meaningful MTF 50% or at least MTF 30% are truly what your eyes will enjoy   That's main reason why DXOMark changes to "perceptual sharpness" measurement as most people truly confused by MTF data, which MTF data?

I have some quotes from above two links,

Contrast levels from 100% to 2% are illustrated on the right for a variable frequency sine pattern. Contrast is moderately attenuated for MTF = 50% and severely attenuated for MTF = 10%. The 2% pattern is visible only because viewing conditions are favorable: it is surrounded by neutral gray, it is noiseless (grainless), and the display contrast for CRTs and most LCD displays is relatively high. It could easily become invisible under less favorable conditions.

The eye is relatively insensitive to detail at spatial frequencies where MTF is low: 10% or less

This measurement, also called “vanishing resolution”, corresponds to an MTF of roughly 10-20%. Because this is the spatial frequency where image information disappears— where it isn’t visible, and because it is strongly dependent on observer bias, it’s a poor indicator of image sharpness.

-- hide signature --
Reply   Reply with quote   Complain
qianp2k
Senior MemberPosts: 9,527Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to Steen Bay, Apr 13, 2013

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:/

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

I'd suggest everyone reads these articles to understand these concepts first and understand what that means in real world photos.

http://www.normankoren.com/Tutorials/MTF.html

http://www.imatest.com/docs/sharpness/

From above two DXOMark MTF tests, we can see 7D only extends at MTF 15% or below that virtually invisible by eyes in real world photos even at base ISO when 7D photos buried by higher noises/grains. So that 7D higher resolution (18mp vs 12.8mp from 5D) is only on paper or reflected in larger CR2 file size but unless you have an eagle eye, you just cannot see it

A bit of sharpening makes it easier to see. It's clearly visible if comparing the resolution test chart shots in DPR's 5D and 7D reviews.

a) that depends from shooting distance.  A meaningful test should be done from at least 50x focus-length distance; b) depends on which lens. 7D needs a really good prime such as 100L (85/1.8 is also very nice lens) under perfect neutral gray and grainless/noiseless light condition at base ISO 100; c) doesn't reflect in real-world photos in outdoor harsh light or indoor poor light; d) depends on measured in center-only, or also in edges/corners so on average resolution.  5D is especially better in edges/corners even with the best lenses.

More meaningful MTF 50% or at least MTF 30% are truly what your eyes will enjoy   That's main reason why DXOMark changes to "perceptual sharpness" measurement as most people truly confused by MTF data, which MTF data?

I have some quotes from above two links,

Contrast levels from 100% to 2% are illustrated on the right for a variable frequency sine pattern. Contrast is moderately attenuated for MTF = 50% and severely attenuated for MTF = 10%. The 2% pattern is visible only because viewing conditions are favorable: it is surrounded by neutral gray, it is noiseless (grainless), and the display contrast for CRTs and most LCD displays is relatively high. It could easily become invisible under less favorable conditions.

The eye is relatively insensitive to detail at spatial frequencies where MTF is low: 10% or less

This measurement, also called “vanishing resolution”, corresponds to an MTF of roughly 10-20%. Because this is the spatial frequency where image information disappears— where it isn’t visible, and because it is strongly dependent on observer bias, it’s a poor indicator of image sharpness.

-- hide signature --
-- hide signature --
 qianp2k's gear list:qianp2k's gear list
Sony RX100 Canon EOS-1D Mark III Canon EOS 5D Mark II Canon EOS 5D Mark III Sony Alpha 7R +20 more
Reply   Reply with quote   Complain
Steen Bay
Veteran MemberPosts: 6,188
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to qianp2k, Apr 13, 2013

qianp2k wrote:

Steen Bay wrote:

qianp2k wrote:

Steen Bay wrote:/

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

I'd suggest everyone reads these articles to understand these concepts first and understand what that means in real world photos.

http://www.normankoren.com/Tutorials/MTF.html

http://www.imatest.com/docs/sharpness/

From above two DXOMark MTF tests, we can see 7D only extends at MTF 15% or below that virtually invisible by eyes in real world photos even at base ISO when 7D photos buried by higher noises/grains. So that 7D higher resolution (18mp vs 12.8mp from 5D) is only on paper or reflected in larger CR2 file size but unless you have an eagle eye, you just cannot see it

A bit of sharpening makes it easier to see. It's clearly visible if comparing the resolution test chart shots in DPR's 5D and 7D reviews.

a) that depends from shooting distance.  A meaningful test should be done from at least 50x focus-length distance; b) depends on which lens. 7D needs a really good prime such as 100L (85/1.8 is also very nice lens) under perfect neutral gray and grainless/noiseless light condition at base ISO 100; c) doesn't reflect in real-world photos in outdoor harsh light or indoor poor light.

Not 100% sure, but think that DPR uses EF 85/1.8 on Canon FF and EF 50/1.4 on Canon APS-C. Agree that the 5D has better sharpness/contrast/acutance, but the 7D's higher MP count will most often also give it a higher 'resolution' (at least at low ISOs).

Reply   Reply with quote   Complain
qianp2k
Senior MemberPosts: 9,527Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to schmegg, Apr 13, 2013

schmegg wrote:

qianp2k wrote:

The result is clear as we have seen from tons of real world photos from respective cameras.

In terms of resolution, I've not seen that.

Which resolution, MTF 50 or 30 or below 20?  I doubt your eyes can see MTF 20% or below in real world photos where 60D/7D suffers higher grains/noises (that is also measured on SNR ratio that even 8-yr-old 5D is still noticeable higher cross entire range of ISO 100-3200).  MTF 50 or at least MTF 30 resolution ultimately matters to normal human eyes

And you've been unable to show this yourself with your 5D and 60D.

Actually shown well in my test samples between 5D and 60D even with one of the best lenses I used in test, 24-70L II.  I gave 60D an advantage by upsampling 5D file to the same size of 60D for easier comparison (without a wrong perception that bigger size has more resolution).

Please not to repeat words over and over again as we have debated enough in that thread.  The ones shown 60D with 24-70L II has very slightly higher resolution on the dollar bill (in very small center area) exactly shown in that very close distance (macro mode) under perfect light condition, grainless/noiseless, so around MTF 20% I guess.  But I have not shown edges/corners yet where it would show with even 24-70L II, 5D haa higher resolution.  So should we consider AVERAGE frame resolution rather very small center resolution.  With inferior lenses and in real world photos, then 5D leads larger while 7D will be hammered more in higher noises/grains.

http://www.dpreview.com/forums/thread/3447852?page=7

http://www.dpreview.com/forums/thread/3447852

5D + 24-70L II upsampling to the same size of 60D

60D + 24070L II

Though I can easily show an 18MP crop image will resolve more of the detail projected onto it than a 22MP FF.

LOL.  Please put a disclaimer next time when you show your "currency bill" shots side by side with 7D and 5D3 that you shoot from the same distance with the same focus length, then you crop out 5D3 photo to match to 7D AOV in very small center area (currency bill).  In such senario, then I have no dispute as that kind of test is meaningless.  All creditable lab tests are tested under the same AOV for the same scene.  With different AOV from same distance/same length  you ended with two different photos, two different scenes.

If you believe your 18mp 7D resolves more than your 22mp 5D3, you should use 7D in your landscape, studio and portraiture photography.  That's only your choice.  I know I will pickup my 5D cameras over my 60D anytime (except from operation consideration when I want to carry smaller/lighter 60D) in those areas of photography 

Why do you think this is? Could it possibly be that you don't know how to read those lines properly I wonder? For instance, did you realise that both those graphs show that the crop resolves more than the FF but with less acutance? I guess not.

Please understand these articles to understand MTF resolution.  BTW, your 22mp 5D3 has higher MTF resolution than 18mp 7D in any MTF percentage from 1-100% as long as frame the scene in the same AOV.

http://www.normankoren.com/Tutorials/MTF.html

http://www.imatest.com/docs/sharpness/

Once again it goes to show the folly of using lens tests to draw conclusions regarding the resolving capabilities of different sensors. All those MTF curves will tell you, when it comes to comparing different sensors, is that lower resolving sensors will give you sharper images from lenses of lesser quality than higher resolving sensors - and that's what you can see with the 85/1.8 - and why the MTF lines are much closer with the 100/2.8.

The tests are measuring acutance and it stands to reason that larger pixels will give increasingly sharper images at pixel level than smaller ones as the lens quality drops. It actually says very little about the relative resolving capabilities of the sensor and a whole lot more about the quality of the lens (surprise, surprise! Who'd have thought that a lens test would be measuring the performance of a lens! LOL!)

Please read and understand.  Both acutance and MTF resolution are related in what human eyes can resolve, the term Perceptual Sharpness DXOMark uses.

BTW - I can now see why DxO don't publish them any more! This is a good example of why they tried to come up with something a little easier for people to understand!

I believe DXOMark P-MPix makes senses to help most people (especially beginners) to understand easier.  Most (think about members in DPR are still minority among all photographers in the world) don't understand MTF data well no mention which MTF resolution, 50, 30, 20, 10?  In real world photos MTF 50 or at least MTF 30 ultimately matter to human eyes.  Under overwhelming most real world scenarios your eyes just cannot see MTF 20% or below or at least not easily.

-- hide signature --
 qianp2k's gear list:qianp2k's gear list
Sony RX100 Canon EOS-1D Mark III Canon EOS 5D Mark II Canon EOS 5D Mark III Sony Alpha 7R +20 more
Reply   Reply with quote   Complain
aftab
Veteran MemberPosts: 7,526Gear list
Like?
Re: Return of Interesting article on DxO about 5D III and D800...
In reply to Just another Canon shooter, Apr 14, 2013

Just another Canon shooter wrote:

Steen Bay wrote:

qianp2k wrote:

With one of the best lenses 100L, we can see both systems are very close but 5D still edges out a little bit.

With a bit lesser lens 85/1.8 (still very good lens), 5D leads wider.  With further inferior lenses, then 5D will further widen the lead.  The result is clear as we have seen from tons of real world photos from respective cameras.

-- hide signature --

We can also see that the 7D actually has a higher resolution. The 5D's MTF curve stops at app. 61 lp/mm because it runs out of vertical pixels (app. 61 lp/mm represents the 5D's Nyquist frequency), while the 7D's MTF curve continues to app. 72 lp/mm (on 24 x36mm). The contrast is rather low, but the 7D's resolution is nevertheless higher.

As I said in another thread a few weeks ago, the extinct resolution of the 7D is higher but that is at a very low contrast. In most of the contrast range, the 5D is either equal (despite having less pixels), or wins easily. I am not sure why this should be surprising to anybody. The  crop penalty is real, and the (low) mp penalty is real, too.

Fair assessment. I was actually expecting the results to be much closer, especially with 85/1.8. As far as I can remember Canon publishes their MTF data measuring 10 and 30 line pairs, 10 being a good indicator of microcontrast and 30 of resolution. I have read somewhere that human eye can resolve up to 36 line pairs per mm. That's a lot if you think about it. For most people it is going to be between 20 and 30. For the argument's sake if we take 60 line pairs and bring MTF down to 20 5D still wins over 7D. What that means in terms of viewing or perception is that 5D images will appear to have more resolution because discernible (to human eye) detail will have more contrast. They probably can be sharpened better with more pleasant results in the post too.

BTW, the real test should be at equivalent FL and equivalent f-stops with common lenses. The 5D will win by even more from MTF20 up, until you get to diffraction limited apertures. The 7D will still be able to show some measurable contrast behind the 5D Nyquist frequency.

When you go to high ISO, even to high equivalent ISOs, the advantage of the 7D will be buried in the noise. On the other hand, the 5D is noisier with equivalent ISOs.

Another observation is that DXO (now) shows the 5D to be clearly the winner in the 85/1.8 case. You can reverse engineer this to make the educated guess that their P-mpix score is weighed more towards medium contrast performance.

I was thinking the same. They probably are measuring 20 to 30 line pairs down to MTF 20.

-- hide signature --

Life is short.
Travel with passion.
http://www.flickr.com/photos/catch45/

Reply   Reply with quote   Complain
MAC
MAC
Forum ProPosts: 12,268Gear list
Like?
true and 14-24 on d800 only scores 17
In reply to Mako2011, Apr 14, 2013

Mako2011 wrote:

Just another Canon shooter wrote:

BTW, the real test should be at equivalent FL and equivalent f-stops with common lenses.

When comparing resolution across formats you have to keep the subject distance constant. DxO apparently doesn't do that. Might be a source of some of their inconsistencies

true !

I find it interesting that: 14-24 on d800 only scores 17 MPix

they haven't tested the 17 T/S on 5d3 yet.  My guess is that it scores 19 like the siggy 35

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

 MAC's gear list:MAC's gear list
Canon EOS 5D Canon EOS 650D Canon EOS 100D Canon EF 70-200mm f/2.8L USM Canon EF 100mm f/2.8L Macro IS USM +5 more
Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads