E5 vs D700--tests

[ptomsu]

Ah yes - because they don't like people to complain about moire I suppose.

They could do the same which MF vendors do or Leica for example.

I shot thousands of images with DMR and M8 and never ever had an issue with moire

But to overcome any possibility that moire appears, they could invest in their processing engines to do same math for eliminating moire

I think this is almost something like ignorance. I never understood why a D3X has to have this thick AA filter. Or their Canon counterparts.

 

[pellicle]

I think this is almost something like ignorance. I never understood why a D3X has to have this thick AA filter. Or their Canon counterparts.

to make the later models look better out of the box (when they reduce the AA filter)?

 

[Paratom]

My thoughts running circles...
Now lets assume 4/3 does have more DOF at same comparable FOV and f-stop than a larger sensor.
That would mean for same (shallow) shallow DOF I need on f-stop faster lens compared to ff.
Does this mean that I would need the f2.0 zooms to get the full felexibility regarding DOF and regarding optical quality?
Or are the f2.8 (-4) lenses as good as lets say the Nikon 24-70?

If one would use the f2.0 Oly lenses they get every bit as big as a full frame camera with the f2.8 pro Nikon lenses:

Oly 14-35/2.0 900g Nikon 24-70/2.8 915g

12-60/2.8-4.0 575g Nikon 24-120/4.0VR 710g

35-100/2.0 1650g Nikon 70-200/2.8 1470g
50-200/2.8-3.5 995g 100-300/4.5-5.6VR ca. 80 x 143,5 745

 

[ptomsu]

I think that the Oly 2.0 lenses are similar weight as the Nikon 2.8 lenses (zooms) - not discussing 100 - 200g more or less of course.

I am pretty sure that Oly lenses are equally good (or even better) than their Nikon (or Canon or Sony/Zeiss) counterparts - even if I open a rathole here.

But where it really comes to play for Oly 43 is the tele range. The Nikon (or Canon) counterparts are for sure much bulkier and heavier. Especially if you go for the Oly 2.8/95-250. Think about that, it replaces a 2.8/200, 2.8/300, 2.8/400 AND 2.8/500 (which I think is not even available for either Nikon or Canon - but I might be wrong). What comes closest is the 4/200-400 from Nikon but this is very limited compared to the Oly lens.

And I think also the "cheaper" standard zoom is not bad WRT IQ and DOF, but pretty cheap for such a great lens.

So I think it depends on where you want to go. If wildlife and extreme telephoto is your preferred area of usage then the E5 concept would win in my eyes. If it is more into available light and highest speed primes (1.4) then it is for sure Canon or Nikon. If it comes to highest speed AF and AF flexibility then it is also Nikon or Canon.

If it is highest resolution - so if you want to use your DSLR also for landscape - then I think it is better to wait for the next Nikon D4X or Canon 1DsMK4 etc, because those will be in the range above 30MP. And then you would not need a separate MFDB system with either 39 or 40MP - I know there are still differences in micro contrast and dynamic range, but they become pretty small actually.

But if you want a great overall DSLR system with excellent and high speed lenses, then the E5 and some of these Zuiko's are the way to go.

For me it gets quite clear the longer I think about this, that I might actually switch to the E5 and some high speed zooms. Plus add maybe the 1.4/53 Sigma with 43 mount in order to have a high speed prime.

 

[mathomas]

Very interesting. I never would have considered an Olympus, or a 4/3 camera, until having read this. I bought a Leica M8 partially because I was tired of the bulk of my Canon 30D and associated super-size lenses. I still want an SLR around for certain situations though, so maybe a swap of the Canon equipment for a basic kit as reviewed here makes sense.

Thanks...

 

[Paratom]

Peter-
I know you are a gearhead and a "switcher" like me.

While the D700 is the most reliable camera in nearly every aspect (specially AF accurancy where I have been unhappy with various other cameras including the 7d and also the 55), I am not totally happy with the overall look of the images. (Skin tones and micro detail).
M9 clear keeper, MF a keeper because I dont get enough money if Iwould seel it.

DSLR for the kids and some sports. So would do I really need and want from my DSLR:
1) fast and reliable AF
2) a nice (optical) viewfinder
3) a good standard zoom, prefered not too big
4) eventually a wide angle zoom (however if 24mm fov of the standard zoom is good this might be enough)
5) a good telezoom
6) a fast 50mm fov lens for low light, prefered f1.4
7) a fast protrait lens with a 65mm-85mm FOV range, prefered would be f1.4
8) later a 150/2.0

In case of E5 this would translate into 12-60, 25/1.4, 45-200, maybe 50 Sima 1.4 for portrait (however I would prefer slightly shorter focal length)

If I had the chance I would like to compare the IQ between the D700 and the E5 myself and see how it works for my taste.

Still - more than 12MP would not hurt at all.

I think that the Oly 2.0 lenses are similar weight as the Nikon 2.8 lenses (zooms) - not discussing 100 - 200g more or less of course.

I am pretty sure that Oly lenses are equally good (or even better) than their Nikon (or Canon or Sony/Zeiss) counterparts - even if I open a rathole here.

But where it really comes to play for Oly 43 is the tele range. The Nikon (or Canon) counterparts are for sure much bulkier and heavier. Especially if you go for the Oly 2.8/95-250. Think about that, it replaces a 2.8/200, 2.8/300, 2.8/400 AND 2.8/500 (which I think is not even available for either Nikon or Canon - but I might be wrong). What comes closest is the 4/200-400 from Nikon but this is very limited compared to the Oly lens.

And I think also the "cheaper" standard zoom is not bad WRT IQ and DOF, but pretty cheap for such a great lens.

So I think it depends on where you want to go. If wildlife and extreme telephoto is your preferred area of usage then the E5 concept would win in my eyes. If it is more into available light and highest speed primes (1.4) then it is for sure Canon or Nikon. If it comes to highest speed AF and AF flexibility then it is also Nikon or Canon.

If it is highest resolution - so if you want to use your DSLR also for landscape - then I think it is better to wait for the next Nikon D4X or Canon 1DsMK4 etc, because those will be in the range above 30MP. And then you would not need a separate MFDB system with either 39 or 40MP - I know there are still differences in micro contrast and dynamic range, but they become pretty small actually.

But if you want a great overall DSLR system with excellent and high speed lenses, then the E5 and some of these Zuiko's are the way to go.

For me it gets quite clear the longer I think about this, that I might actually switch to the E5 and some high speed zooms. Plus add maybe the 1.4/53 Sigma with 43 mount in order to have a high speed prime.

 

[ptomsu]

Peter-
I know you are a gearhead and a "switcher" like me.

While the D700 is the most reliable camera in nearly every aspect (specially AF accurancy where I have been unhappy with various other cameras including the 7d and also the 55), I am not totally happy with the overall look of the images. (Skin tones and micro detail).
M9 clear keeper, MF a keeper because I dont get enough money if Iwould seel it.

DSLR for the kids and some sports. So would do I really need and want from my DSLR:
1) fast and reliable AF
2) a nice (optical) viewfinder
3) a good standard zoom, prefered not too big
4) eventually a wide angle zoom (however if 24mm fov of the standard zoom is good this might be enough)
5) a good telezoom
6) a fast 50mm fov lens for low light, prefered f1.4
7) a fast protrait lens with a 65mm-85mm FOV range, prefered would be f1.4
8) later a 150/2.0

In case of E5 this would translate into 12-60, 25/1.4, 45-200, maybe 50 Sima 1.4 for portrait (however I would prefer slightly shorter focal length)

If I had the chance I would like to compare the IQ between the D700 and the E5 myself and see how it works for my taste.

Still - more than 12MP would not hurt at all.

Well, yes, I think I am a gearhead. Committing this once, gives you immediately much more freedom talking about gear and switching between different systems

Anyway, I also would like to do a side by side comparison of D700 and E5. For my purpose I would do it with 2.8/24-70 Nikkor on D700 and 2/14-35 Zuiko on Olympus, that would be perfect. Not sure I can achieve that.

WRT to fast primes on the E5, I would prefer the 1.4/35 Sigma, I owned the 1.4/25 on my E3 and was absolutely unhappy - maybe it was only my sample, but I also did not like the still very noticeable wide angle character of this lens. So for me it was never kind of a standard lens replacement.

What I fear is wether Olympus is really committed to 43 or they kind of will abandon and only go M43 in the future. While that is not as bad because of the already good quality of the E5, As you said I would love to have some more MP in the future in a native 43 body and not in a M43 body with adapters.

 

[Paratom]

It will be still interesting to see what comes next from Nikon- since this comparison is a 2-3 year old D700 vs a brand new E5.

Well, yes, I think I am a gearhead. Committing this once, gives you immediately much more freedom talking about gear and switching between different systems

Anyway, I also would like to do a side by side comparison of D700 and E5. For my purpose I would do it with 2.8/24-70 Nikkor on D700 and 2/14-35 Zuiko on Olympus, that would be perfect. Not sure I can achieve that.

WRT to fast primes on the E5, I would prefer the 1.4/35 Sigma, I owned the 1.4/25 on my E3 and was absolutely unhappy - maybe it was only my sample, but I also did not like the still very noticeable wide angle character of this lens. So for me it was never kind of a standard lens replacement.

What I fear is wether Olympus is really committed to 43 or they kind of will abandon and only go M43 in the future. While that is not as bad because of the already good quality of the E5, As you said I would love to have some more MP in the future in a native 43 body and not in a M43 body with adapters.

 

[ptomsu]

It will be still interesting to see what comes next from Nikon- since this comparison is a 2-3 year old D700 vs a brand new E5.

Right, any ideas, rumors?

I could not find anything substantial so far .....

 

[Paratom]

Right, any ideas, rumors?

I could not find anything substantial so far .....

no - I have no idea.

I would wish for more MP, weaker AA-filter (my main wish), and slightly different colors but I doubt this will happen.

Sorry for being off topic. I will stop here

 

[Godfrey]

It will be still interesting to see what comes next from Nikon- since this comparison is a 2-3 year old D700 vs a brand new E5.

The logical "next step" for the Nikon D700 is to incorporate video capture, as Canon did with the 5D to 5D II model upgrade (and Nikon did with the D300 to D300s too). Whether they up the pixel resolution or improve anything else ... yeah, well, I bet they'll do something.

IMO, All these high-end cameras are reaching the point of diminishing returns in pixel density and low light capability. The next great avenue of development is almost certainly video capabilities ... all of them have a lot of growth in that direction yet to come.

I'm not a "switcher" although I'm brand agnostic (except for Sigma ... won't buy anything from them at all . I participate in the gear-head-curiosity game but keep it separate from my work in Photography ...

... "Equipment often gets in the way of Photography." ...

 

[nugat]

Still - more than 12MP would not hurt at all.

It COULD hurt the picture quality.
Remember the Mhz race? At ca 3Ghz it turned out faster PCs deliver diminishing returns. Now it's multithreading, memory, bus speed etc.
Really good 12Mpix can print up to A2 size undistinguishable from 24MPix originating file*. How come?
First you have the imaging system resolution. The product of : lens, AA filter, sensor, processor, debayer and JPEG/raw developer software. 4/3 was optimized for digital and nobody knows it better than Olympus. The result maximum horizontal resolutions are 2350 lw/ph or line widths per picture height (GH1 +Zuiko 50/2) and 2700 lw/ph for all FF 24Mpix combos. The difference is of 15% at most. Or used to be, as I believe E5 with the 14-35/2 could be at 2500 lw/ph (waiting for that glass).
It seems to me that E5 is perfectly tuned to Zuiko glass and there is little to be gained from more pixels ( first reports on GH2 18 Mpix).
What we need more now is dynamic range, color, low noise. To me the megapixel race might as well stop at 12 million pixels for printouts.
For electronic display it stopped years ago. Who shows their pictures on 4000x3000 displays? Not too many of those around.
*cropping to A# (sq root 2) print format is equally depriving for 3:2 and 4:3 ratios.

 

[ptomsu]

It COULD hurt the picture quality.
Remember the Mhz race? At ca 3Ghz it turned out faster PCs deliver diminishing returns. Now it's multithreading, memory, bus speed etc.
Really good 12Mpix can print up to A2 size undistinguishable from 24MPix originating file*. How come?
First you have the imaging system resolution. The product of : lens, AA filter, sensor, processor, debayer and JPEG/raw developer software. 4/3 was optimized for digital and nobody knows it better than Olympus. The result maximum horizontal resolutions are 2350 lw/ph or line widths per picture height (GH1 +Zuiko 50/2) and 2700 lw/ph for all FF 24Mpix combos. The difference is of 15% at most. Or used to be, as I believe E5 with the 14-35/2 could be at 2500 lw/ph (waiting for that glass).
It seems to me that E5 is perfectly tuned to Zuiko glass and there is little to be gained from more pixels ( first reports on GH2 18 Mpix).
What we need more now is dynamic range, color, low noise. To me the megapixel race might as well stop at 12 million pixels for printouts.
For electronic display it stopped years ago. Who shows their pictures on 4000x3000 displays? Not too many of those around.
*cropping to A# (sq root 2) print format is equally depriving for 3:2 and 4:3 ratios.

Very well said!

Nevertheless, what do you think then about the rumored next generation Pro bodies from Canon and Nikon which should offer resolutions between 30-40MP? Assume that 40MP would be possible, how would that translate into lw/ph (if we assume that their lenses can follow this) ?

And let me add the (more rhetorical) question: Why do camera vendors still sell and advertise their cameras (systems) with MP count and not better with lw/ph count? I think I know some answer here, but would like to hear your version ....

Thanks!

 

[Paratom]

the d3x/a900/5dII do show that with good glass higher MP than the d700´s 12MP does not hurt IQ at low and medium ISO.
Upscaling images makes me allways feel like adding information which has not been there in reality.

 

[ptomsu]

the d3x/a900/5dII do show that with good glass higher MP than the d700´s 12MP does not hurt IQ at low and medium ISO.
Upscaling images makes me allways feel like adding information which has not been there in reality.

Hmm - not saying myself that higher resolutions hurts, but I owned the 5D2 and the A900 and I always own the best glass for these systems (I know I am a gearhead ) but I must say that I was not so impressed with the outcome of both cameras - A900 of course was better, but resolution wise and detail wise (I used Zeiss/Sony glass) I was not overwhelmed.

I think part of this is the strong AA filters of these cameras. So I also think in order that the next step in resolution makes sense the AA filters have to get MUCH weaker or disappear totally. Like in MFDBs.

Of course adding some 5-6MP to the current 12MP E5 sensor would bring some benefits and I am pretty sure that Olympus could handle the (fast) processing of these date. I simply believe that at the time they started the design/development of the E5 there was no such 18MP 43 sensor, as it will show up in the GH2. And their marketing department said they need to come NOW with the E5 and not some 6 months later ....

 

[nugat]

Very well said!

Nevertheless, what do you think then about the rumored next generation Pro bodies from Canon and Nikon which should offer resolutions between 30-40MP? Assume that 40MP would be possible, how would that translate into lw/ph (if we assume that their lenses can follow this) ?

And let me add the (more rhetorical) question: Why do camera vendors still sell and advertise their cameras (systems) with MP count and not better with lw/ph count? I think I know some answer here, but would like to hear your version ....

Thanks!

I am not an engineer, I just play one in the commercial.
Advertising in Megapixels is similar to horsepower in cars. The real test is the time per one loop of Nurburgring racetrack. Or lw/ph of the system.

30-40MPix is, say, a 7500x5000 pixels. The Nyquist limit of 5000 pixels tall sensor is 2500lp/ph (line pairs per picture height) or 2500/24mm=104 lp/mm (line pairs per milimeter). The current top FF lens deliver 40-50 lp/mm at 50% contrast (MTF). To deliver 100lp/mm they would need to be 2-2.5x bigger. Good luck.
PS. Olympus zuiko SHG is as big as FF lens just to deliver the 100lp/mm resolution MTF50 on the 4/3 sensor. 13mm x100= 1300lp/ph whereas the theoretical Nyquist limit is 1500 lp/ph . The actual performance at 85% Nyquist is world record along with Leica M9+Elmar Leica glass (Erwin Puts at imx.nl). A good figure is 70% of Nyquist.

 

[ptomsu]

@ nugat

Well I did a little calculation myself and I come to the following:

E5 with a sensor size of 13 mm ph x 17.3 mm pw with 12.3 MP results in 225 mm2 and this means 54666 pixels/mm2 which results in approximately 238 pixels/mm. Given a ph of 13mm these are 3094 pixels. With Nyquist you end up in 1547 lp/ph - same result as you came up.

If I go to the Oly website, I only can find MTF for e.g. SWD 14-35 which shows a highest contrast value of approx. 95 with 10 lp/mm or 75 with 60 lp/mm at Wide end. From this I cannot see how you come to 100 lp/mm at 50 percent contrast - is this value true???? If so then your calculation is right that 100 lp/mm x 13mm results in 1300 lp/ph.

Now if I do the same for D3X FF ....

D3X with a sensor size of 24mm ph x 36 mm pw with 24.5 MP results in 864 mm2 and this means 28356 pixels/mm2 which results in approximately 168 pixels/mm (which BTW is only 70% of the pixel density of the E5, so there is for sure room for improvement, as long as Nikkor lenses add to this). Given a ph of 24mm in FF you end up in 4032 pixels. With Nyquist you end up in 2016 lp/ph.

Looking at the Nikon website, for the 2.8/24-70 for MTF you find a highest contrast value of approx. 98 with 10 lp/mm or 88 with 30 lp/mm. If I do a similar extrapolation as you assumed for Zuiko at 50% contrast value, this should at least also result in 100 lp/mm. This gives you then 2400 lp/ph - slightly above the Nyquist limit of this 24.5 MP sensor.

What can we conclude from of this:

1) Top Zuiko glass and top Nikkor glass are equally good - at least WRT MTF!
2) With the E5 and 12.3 MP you already reach (or come slightly over) the theoretical limit of 1300 lp/ph for a top Zuiko lens. So even a higher resolution sensor would not add to this.
3) With Nikkor and D3X you are still below that limit. Which means you could have a 34.5 MP FF sensor, (2400x2 by 2400x3 results in 34.4MP) in order to make full use of state of the art top Nikkors.

I think what we have so far left out of these considerations is the AA filter and post processing. The strong AA filter in the D3X for sure will limit that theoretical 2016 Nyquist limit to maybe 1700 - 1800. Given that it would be fair to compare the allow to compare the E5 with its 12.3 MP resolution to the D3X with 24.5 MP resolution, always assuming one uses top lenses of both brands.

Question remains - will Olympus be able to further improve with higher resolution sensors (say 18MP)? And how far better will a D4X with 34.5 MP work with the top Nikkor lenses.

MTF Charts attached - first 2 are Zuiko at 2.0 at 14 and 35 respectively, second 2 are Nikkor at 2.8 at 24 and 70 respectively.

 

[nugat]

I should not have been getting onto the murky ground of resolution in the first place.
First of all I know nothing how manufacturers MTF graphs are obtained. If somebody can point me to a statement from a major maker (CaNikon etc) how they do that, I'd be grateful. Are they theoretical figures? Measured? How? On a camera? With optical bench/ equipment?
Additionally they present their figures in often incomparable formats: different lines numbers is common (10/30 for Canikon, 20/40 for Panasonic, 20/60 for Olympus etc etc).
Secondly all review sites I know test whole systems (lens, AA filter, sensor, electronics, algorithms) and not lenses. Moreover almost all of them use Imatest software, often which they use incorrectly (I have examples but have no intention to wage private wars here).
The only site I have some trust on methodologies is Erwin Puts at imx.nl.
What we would really need to see for a meaningful comparison of LENS parameters is a comprehensive set of scientific methods (lens projector, optical bench with MTF analysis etc). Maybe it's an idea for a business.
Having said that, I have to admit that my figures are conjectures based on anecdotal and circumstantial evidence rather than scientific facts.
First is a thorough reading of the aforementioned Puts site and what he has to say about lenses.
Second is cinematography. Here glass is tested with lens projectors mostly. We see that Zeiss Master series is really better than others. We also know that Arriscan considers 4k scanning as sufficient for resolution and 6k for other aspects of picture information. That roughly corresponds to 4000dpi used in Nikon film (36x24) scanners or the height of the 24 megapixel sensor (6000x4000)
If Arri and Nikon limit their resolution capture at that figure, it's a hint where the actual glass resolution limits lie. In case of arri we talk about 15k a pop cinematography primes, each 2 kg heavy and really huge.
Therefore I do not believe that the current crop FF glass can benefit from a 30-40 mpix 36x24mm sensors. That was also confirmed by Leica starting the S2 line in medium format (37mpix) with a whole new lines of lenses (fairly big). If they thought anything was to be gained in 36x24, the R line would have been continued.
But most of that is conjectures as I said, so better wait for those 30Mpix+ sensors from Canikon and see.

 

[ptomsu]

I think we are not so far from reality with your calculations and mine (or maybe better call it observations).

What the main outcome is that 43 glass at its best can resolve up to 1300 lp/ph if you assume some 50% contrast and 100lp/mm - which can be basically read out of the MTF charts we have and I trust they are somehow correct, otherwise to hell with those companies.

That means the current 12MP sensor is already above the limit what good Zuiko lenses can resolve. Does that mean that higher resolution is not necessary? Potentially not, because you can gain still more lp/ph with higher contrast, so maybe the sweetspot should be around 16-16MP for 43 (and same for M43). Again, not really scientific, but I think we are pretty close to the truth.

For Nikon that measn, that their top grade glass is obviously not worse than Zuiko glass, which is often consensus unfortunately. With their new nanocoated lenses they usually to or are at least on par with Zuiko lenses. What that means in terms of resolution is, that they deliver 2400 lp/ph for FF, which is not exactly double the number which Zuiko would deliver (they would come to 2600 lp/ph) but this is rather theoretical, as Zuiko is not calculates for FF (24 mm ph) but only for 43 (13 mm ph). So for me this is actually quite the same quality!

Now take the current top Nikon sensor in the D3X with 24.5 MP, which can resolve due to Nyquist 2016 lp/ph, which is still a lot below the resolution limit of 2400 lp/ph for top level Nikkors. So this leaves room for improvement. As I said a sensor making use of the full Nikkor resolution would then be 34.5MP, but same is true here, more resolution will not hurt, because this is all calculated for 50% contrast, so at higher contrast levels maybe 40MP would be kind of sweetspot here in FF.

What does that mean? Olympus is currently already at the edge of what makes sense for their 43 system, Nikon has much more room for improvement. Magical? Not at all - simple physics, 4x larger sensor etc. etc.

On the other side I am well aware of other issues you get with higher resolution like diffraction etc, but this is a different story and even more true for Olympus with their smaller sensor size. So has anybody done any meaningful diffraction tests with some top level Zuikos stopped down?

I for myself am very grateful that you initiated this discussion, because it made me investigate myself and start calculating myself and find that I will be better suited to stay with Nikon. But everyone will have their preferences of course

 

[Jerry_R]

Soon we will have more comparisons between existing u43 sensors and new one - coming with GH2, which has increased resolution.
Then we will see if more details can be seen and with which lenses.