Pro-photographers, time to reconsider

[cosmonaut]

Cosmo,

How long have you been using E-M5?
Do share with us the moments (as in which photo and what date) when the camera didn't deliver the quality expected by those magazines you mentioned.

Last time I check, people just *got started* evaluating that one.
Which is why I think the article I refer to is relevant.

As for battery life? get a grip, or two.

My E-M5 is not here yet but have the NEX7. I tried to shoot my daughters soccer games with it and yes it can be done. But the battery died half way thru the game and nailing shots was a PITA. My a77 and a850 blew it off the planet. Just with the CZ glass I have. If I can't shoot a 7 year olds soccer game in day time light a high school football or basketball game in darker light would be a joke.
I am sure there are pros that can make anything work. If a pros only system is a mirrorless camera he or she is going to be handicapped in many situations where if a good DSLR would serve them with everything they do. There is a NASCAR race on tonight at Richmond. Flip it on and look behind pit wall and count how many m4/3rd cameras you see and call me back.

 

[cosmonaut]

It's not the sensor size that's the problem with m4/3rd an Olympus E5 could work for sports. It's performace. The E-M5 maybe the game changer.
But I doubt Canon and Nikon are worried they will lose a share of the Pro market because of the E-M5

 

[semilog]

Are you saying it's physically impossible to implement backside illumination as sensor size increases?

If you can't buy it, then it's because the vendors are stupid... or maybe the cost:benefit ratio decreases as sensor size increases because of the larger sensor's inherent advantage.

But it dosn't matter. A backside large surface area sensor will collects more data than a backside sensor with a smaller surface area.

1. A BSI sensor does not collect "more data." It simply has higher quantum efficiency. Not the same thing. There are technical challenges in making large BSI sensors. For what it's worth, I'm about to buy a second monochrome BSI-EMCCD camera for my laboratory. The chip is 5mm x 5mm, 512x512 pixels (0.25 Mpix). That will be $30,000 for the bare sensor in a box with a Peltier device to cool it to -80° C and a minimal interface. >90% quantum efficiency and capable of operating in single photon counting mode. It will not collect "more data." What it will do is give high SNR under highly specialized conditions. The point being that different sensors are optimized for different applications.

2. If the sensor it is not available, it does not collect any data at all. 24x36mm BSI sensors are not currently available. Such things might be available if you are a military contractor.

3. Cell phone sensors are currently at least two full technology generations ahead of FF camera sensors. Expect that gap to widen, not shrink.

4. This has nothing whatsoever to do with the vendors being "stupid." It has everything to do with cost efficiencies of manufacturing and cost-benefit tradeoffs.

 

[Trius]

You will also never get the equivalent DOF with a m4/3 sensor that you can get with a full frame (36mm x 24mm) sensor. That's my main reason for shooting FF.

Best,
-Tim

And my main reason for shooting LF. 😛

 

[x-ray]

The debate around full frame is the same as when tech pan came out. Yes it was sharp but it didn't have the same tonality and gradation as larger formats. Ok 4/3 and APSC are sharp and look great but they aren't as smooth in tonality as larger full frame sensor cameras. The same is true of FF vs MF digital. Sensor size does make a difference. Each time you increase the sensor size it equates to increasing film size. Tonality gets smoother and sharpness is still there and more and more subtle tones and gradations emerge. There's also a totally different presence, feel to the images from a large sensor.

 

[semilog]

Those are mildly interesting assertions. They would be vastly more interesting if backed up by careful comparisons of well-made prints from files shot side-by-side under well-controlled conditions.

 

[Trius]

Carefully controlled tests are useless. I could say the same thing about 35mm film vs medium format or larger. I spent part of yesterday afternoon in the presence of nine Magnum greats, many of whom shoot film/35mm ... Larry Towell, Bruce Gilden, etc. I don't disdain or reject their work because it's made on "miniature" film. Some of the others (Webb, Soth, Wylie) shoot digital either exclusively or as well as film. I like their work, too.

 

[semilog]

Carefully controlled tests can be extremely useful. It depends what you want to know.

But as you suggest, "image quality" is not necessarily a prerequisite for quality images.

 

[semilog]

Ctein has tested almost every popular pro camera image file. He also is up on MFT as he uses one for much of his work. Of the 4 or 5 world class Dye Transfer printers in the US, he was at or near the top. Any questions I have about image file quality are directed to Ctein.

Exactamundo. I have a few of Ctein's prints (including a couple of DT's), and the new one — taken with MFT and printed for the specific purpose of showing what's possible with MFT — is on the way.

 

[ktran]

My E-M5 is not here yet but have the NEX7. I tried to shoot my daughters soccer games with it and yes it can be done. But the battery died half way thru the game and nailing shots was a PITA. My a77 and a850 blew it off the planet. Just with the CZ glass I have. If I can't shoot a 7 year olds soccer game in day time light a high school football or basketball game in darker light would be a joke.
I am sure there are pros that can make anything work. If a pros only system is a mirrorless camera he or she is going to be handicapped in many situations where if a good DSLR would serve them with everything they do. There is a NASCAR race on tonight at Richmond. Flip it on and look behind pit wall and count how many m4/3rd cameras you see and call me back.

I don't know if this is normal, but I shot a half-marathon today with an E-PL2, brought three batteries, and only depleted one of them... after about 1500+ shots. Mind you, these were JPG normal (as per the assignment requirements: we were trying to get as many shots of as many of the thousands of participants as possible), shot in continuous mode. I turned off the camera between bursts and position changes along the route (it's as fast as or faster to start up cold than from "powersave mode"). The CIPA rating for the E-PL2 is 290 shots per charge. Obviously my experience is not typical. I wonder if it's because of how I went about it, plus maybe the fact that I had the IS turned off (when you're shooting wide angle lenses right into the runners' faces at 1/1000s, you don't need no stinkin' IS).

Anyhow, I'm just saying that there may be ways to get around your battery life issue. Certainly, you won't match the battery life of a DSLR due to design philosophy, but I think you can get through a soccer match (I've shot those too).

And as for your earlier post re: the Pentax pancake lenses: please remember that those have to mount on a camera with a much larger lens-to-sensor distance than the m4/3 cameras. I'm not certain the Pentax K01 with that lens is any slimmer than a PEN E-PM1 with the 17mm pancake.

 

[jarski]

wonder where "legacy" suddenly appeared to describe lenses used on full frame sensor bodies. I doubt many pro stick 40 year old Nikkor to his/her D4 when going to assignment. of course he/she could, and use the lens as it was meant to be. which is the best part 🙂

edit, btw. Dpreview has nice tool called Studio Shot Comparison. its possible to pixel peep 100% crops and compare side-by-side latest cameras. scene is well light, which (IMHO) gives little leeway for smaller sensors regarding light shifts and how details in shadows are recorded. anyway good source of information for judging by self, than believe some biased blogger/net commentator.

 

[Trius]

Carefully controlled tests can be extremely useful. It depends what you want to know.

But as you suggest, "image quality" is not necessarily a prerequisite for quality images.

Well, in the context of this thread, they are not very useful.

And yes, "image quality" and "quality image" are two very different things, which gets confused almost every minute on every photo forum past, present, future. 🙄

 

[willie_901]

It's All About Surface Area (rignt now anyway)

It's All About Surface Area (rignt now anyway)

......
1. A BSI sensor does not collect "more data." It simply has higher quantum efficiency. Not the same thing. There are technical challenges in making large BSI sensors. For what it's worth, I'm about to buy a second monochrome BSI-EMCCD camera for my laboratory. The chip is 5mm x 5mm, 512x512 pixels (0.25 Mpix). That will be $30,000 for the bare sensor in a box with a Peltier device to cool it to -80° C and a minimal interface. >90% quantum efficiency and capable of operating in single photon counting mode. It will not collect "more data." What it will do is give high SNR under highly specialized conditions. The point being that different sensors are optimized for different applications.

.......

3. Cell phone sensors are currently at least two full technology generations ahead of FF camera sensors. Expect that gap to widen, not shrink.

Thanks for catching my imprecise use of the word data.

I should have said: as the sensor area increases the information content of the photograph increases.

Data = Signal + Noise

The signal is what we want. It represents a state of nature, or the actual but unknown flow of electrons from each sesnor site. The signal electron flow is proportional to number of photons captured by the sesnor.

The noise is responsible for uncertainty in the data. The are two main sources: quantum noise and read noise (noise floor). One is an inherent property of matter and the other is generated by the camera's electronics.

Quantum efficiency is important. However QE alone is meaningless. If the read noise is high, QE is compromised. The storage capacity of a sensor site is called the saturation capacity. QE is compromised if the electron flow is not proportional to the photons captured by the sensor. This lower the saturation capacity, the less important QE becomes.

Large sensor areas increase the signal level. An increase in sensor area does not necessarily increase read noise. The data from an APS-C sensor has less uncertainty than data from a m4/3 sensor because there is more signal.

There is not more data, but there is more information in the data.

The site

http://www.sensorgen.info/

computes QE, minimum read noise, and maximum saturation capacity for dozens of digital cameras. I have reproduced a few of their results below.

Camera QE Read Noise Saturation Capacity

Pen_E-P3 41% 8.1 17791
Pen_E-PL1 42% 11.2 17424
XZ-1 35% 2.6 6498

DMC-G1 33% 5.9 14346
DMC-GH1 50% 4.2 18662
DMC_G3 45% 2.9 13612
DMC_GH2 43% 3.0 11803
DMC_GX1 44% 2.7 12554

D700 38% 5.3 58111
D7000 48% 2.5 49058

Because efficient exposure maximizes saturation capacity and has no effect on the read noise, the signal (electron flow) is larger for sesnors with more area. The data from a well-designed APS-C camera contains more information than the data from a well-designed m4/3 camera.

I apologize for such a nerdy post. However at this point in digital photography sensor area is important. I don't understand why m4/3 proponents can't just admit they prefer the increase in convenience instead of more information content. After all, for many photographs the reduction in information content has negligible impact on image quality. We all know the image quality if just one factor in the aesthetics of the final image.



Also, who cares about cell phone sensors? The OP suggested people whose livelihood depends, to some extent, on image quality are missing the boat by ignoring m4/3 cameras. I disagreed because convenience does not trump information content for many working photographers.

 

[semilog]

QE is compromised if the electron flow is not proportional to the photons captured by the sensor.

That is not generally an issue. CCD and CMOS sensors are highly linear devices. The bigger issue is that too few photoelectrons accumulate under low-light conditions, limiting the SNR to ≤ sqrt[# of events accumulated]. Higher QE allows detection of more events per photosite.

This lower the saturation capacity, the less important QE becomes.

That simply is not so.

Better quantum efficiency is always desirable. Cell phone sensors are using BSI to compensate for their small photosite area (typically ~1.5 µm pixel spacing with a smaller photosite). BSI gives about a full stop of improvement in real applications. The camera I'm about to buy for my lab (>10 µm pixel spacing) is also BSI, has >90% QE, a full-well capacity >100,000 e-, and read noise of <1e-. That's 10-year-old tech, by the way, and the most recent cameras are not meaningfully better.

If it were economically feasible, all sensors would be BSI. Even 4/3 BSI sensors are still too hard to manufacture and too expensive to put in consumer devices.

Large sensor areas increase the signal level. An increase in sensor area does not necessarily increase read noise.

All else being equal, yes. But all else is seldom equal, which is why the APS-C Pentax K-x was giving the D-700 a run for its money.

The data from an APS-C sensor has less uncertainty than data from a m4/3 sensor because there is more signal.

Again, all else being equal. But are you talking about Nikon APS-C (1.5x crop) or Canon (1.6)? 😛

More seriously, consider the Pentax K-x again. It used a Sony sensor that marked first appearance of many technologies used for the D-7000 and now the D-800. It absolutely crushed every other APS-C sensor on the market when it came out. The problem faced by micro 4/3 cameras has not been sensor size, which is not a lot smaller than APS-C. It's been that the Kodak and Panasonic sensors in 4/3 and micro 4/3 cameras have lagged about a full generation behind what Canon — and especially Sony — were bringing to market.

The sensors in the GH2 and EM5 perform very well, suggesting that that gap is narrowing.

Also, who cares about cell phone sensors?

Anyone who cares about digital imaging should care about cell phone sensors. They represent the state of the art in consumer devices and are where the vast majority of the R & D money is going.

 

[semilog]

The popular cooling method involves Peltier Cooling.

Yes. The microscope camera that I alluded to above runs at -80° C.

That Sony is achieving read noises of <2 photoelectrons at room temperature is nothing short of astonishing. It's nice to know that Mr. Morita's once-unrivalled company is still good at something.

Of course, read noise is not the only noise that designers contend with. There is dark current (helped a lot by cooling) and pattern noise, too. And at small numbers of events, shot noise -- the main reason QE is so important.

 

[hatidua]

Irrespective of all the tech-speak, why do so many people see it as their calling in life to convince others of the virtues of the 4/3 cameras? (or FF sensors, or medium format over 35mm, Etc).

As the article seems to at least slightly be portraying a professional setting, it bears mentioning that I have yet to meet a single competent pro who needs a blog to tell him/her what gear to use in order to satisfy his/her clients needs. Nor, have any of my clients cared what I use to get the results they are paying for, they simply want the image/s. The author acts as if he was hesitant to immediately let his clients in on what tool he used for the job (as if it would matter...).

If someone wants a 4/3 camera, fantastic - go buy one. They are not appropriate for the needs of some and no amount of blogging on their merits is going to change that.

 

[semilog]

Irrespective of all the tech-speak, why do so many people see it as their calling in life to convince others of the virtues of the 4/3 cameras? (or FF sensors, or medium format over 35mm, Etc).

People like to argue. It can be fun and it can be a good way to test your own ideas and knowledge, and learn things.

 

[jarski]

finished browsing the DP's review of the OM-D. conclusion seemed to be that its still an enthusiast camera, and very good one at that, but not pro. no doubt pro's too find many uses for it in their toolkit as well.

"Overall score 80%
The Olympus OM-D E-M5 is certainly the most capable Micro Four Thirds cameras we've reviewed and arguably the most likeable mirrorless model yet. It falls down a little bit on its continuous focusing but we have no absolutely no complaints about the image quality. It's small, attractive and a pleasure to use, and its pictures are equally enjoyable."

 

[dreilly]

The only camera I ever used that had a truly workable continuous AF was the Nikon D300/D700...the 3d focus tracking was something else. On any other camera is was worthless or near it, so I can't hold this against the OM-D, and I think DPreview says as much in the review. What's missing from this being the "pro" Pen Olympus talked about? Maybe only that it's an OM-D and not a Pen.

 

[cosmonaut]

Well like I said Olympus camera sensor DXO marks haven't made any improvement in years. Olympus thins the AA filter and puts a new engine in it for better IQ. This is smoke and mirrors. Other companies are improving on their sensors with every upgrade. Olympus puts out to many models to quickly. They need to slow down and put some R&D into their cameras.
Olympus faithful look as the next camera as the one that will put them in the game only to be let down in the long run. I really wonder has Olympus squeezed as much out of their sensor as they can. From what I am reading this one may be no different than the others. I hope I am wrong.
The EVF of the NEX 7 has twice the resolution of the Olympus EFV and the NEX 7 has been out awhile and was delayed in release as it were. Olympus may never catch up.