Canon 5D MKII
A Field Review by Nick Devlin
The 5D Mark II is a hard camera to review. On one hand, it betters its predecessors in every respect and pushes the envelope of 35mm photography to new bounds at a relatively accessible price. On the other hand, it is not the photographic second coming that its pre-release hype and anticipation suggested. Given that the original 5D brought full-frame to the masses, the messianic metaphor isn’t altogether misplaced. With this camera, Canon offers us mere mortals elevation into the exalted air of 20+ Megapixels – previously the terrain of MF backs and the über-pricey 1DsIII. This is heady stuff, even in the era of camera-a-minute advancements and galloping Megapixel counts.
(In fairness, the Sony A900 had actually arrived offering a higher pixel count and higher frame rate at roughly the same price almost a year earlier but, curiously, no one appeared to notice or care. )
In a lovely symmetry, the 5D’s second-coming was timed perfectly to coincide with my second-going to Antarctica, providing an ideal opportunity to give the camera a long, intense and diverse work-over. Since one of the recurrent themes surrounding the camera’s introduction was the notion that it transcended limitations in resolution and ISO, I made a conscious decision to shoot with it as it there were no limits. Whatever the light or the ISO, we were going for it. And going for it with an expectation to print BIG.
What follows below is my rather subjective evaluation of the 5D2’s characteristics which matter to me, based on my first three months of living with this camera. For the impatient amongst you, here’s the gist of it: while the 5D2 is not the holy grail, it is a defining camera. While it is not the camera to end cameras, nor even necessarily the best camera in its own class, the limitations of photography have been pushed back once more, to limits we probably could not have imagined a mere decade or two ago.
Let’s start at the beginning. Coming out of the box, the 5DII looks, feels and smells like one of the family. It delivers Canon’s tried-and-tested EOS ethos and interface. There’s really nothing new here, save for the lusciously large rear LCD and the selector stick, which is becoming standard on most upper-tier DSLRs. This is still a ‘modal’ camera, driven by multiple button-pressing, wheel spinning, domaflicky toggling nonsense. But this is sad fate of our entire discipline, so its useless to savage Canon for continuing to embrace this unfortunate aspect of the photographic zeitgeist.
The mode dial on the top left of the camera is much better that the 1-series’ even deeper modal morass, and is one of my favourite features of the 5DII. Unfortunately, the dial has a habit of getting accidentally turned when the camera is tossed over one’s shoulder, resulting in unintended and shot-wrecking mode changes. I don’t fully understand this, since the dial-tension feels just about right to the touch.
As for size, the 5DII proves that bigger is not always better, especially where cameras are concerned. While the 1-Series EOS cameras have always been a joy to hold, they are heavy beasts, unsuitable for a lot of applications where size and weight matter. The 5DII finally puts a top-tier chip into a smaller-sized camera body, which is all good. Naturally, the first thing I did was run out and put a grip on it. Why? Well, taller cameras simply balance better when long lenses are mounted. Also, the grip permits use of the hand-strap, I find to be an invaluable accessory. While it increase the camera’s size, the grip adds relatively little weight, especially with only one battery installed.
"Penguin Blues", Desolation Island, January 2009
Compared to the professional-level 1-Series cameras, the 5DII has a softer and quieter shutter release. For a relatively fast, full-frame DSLR it is quite discrete. That said, I did really enjoyed the ‘solidity’ of the 1DsIII which also made the trip with me. There is a seriousness of build evident in the 1-Series which the 5DII lacks. It’s kind of the like the difference between a police-issue Ford Crown Victoria and a Chevy Lumina. One feels like you could drive it through a brick wall. The other doesn’t.
None of this speaks to the ultimate durability of the 5DII. While a relatively large number of them developed problems on the Antarctic trip, this is a function of weather-sealing (or the lack thereof). With one exception, all of the 5DIIs that died did so after being exposed to conditions in which a failure should not have been surprising. I have no doubt that scores of pros will take millions of problem-free frames with 5DIIs in the years to come.
Personally, I found the camera fast enough for anything but serious sports photography. The buffer is big, write-speed to UDMA cards is very quick, and the frame-rate it fully adequate for everything but ultra-fast action. To add a comparative perspective, however, the Sony A900 is a lot faster, without feeling like it is. The Sony employs some form of mirror-damping technology which is effective in making it fly effortlessly at 5fps, while pumping out a veritable monsoon of data.
Beyond frame-rates and data-push, the 5DII is faster than previous generations of EOS cameras in all of those little intangible ways that make a machine either pleasurable or pedantic to use. Its just like that enthralling little speed-kick you get every time you upgrade your computer. This is a fast, smooth mature piece of technology.
Though I have heard griping about the 5DII’s AF performance, I found these unsubstantiated in a real-world setting. It’s not a blazingly fast as the top-tier Nikons, but proved more than up to the task of shooting anything we encountered in a nature/wildlife shoot, including a passable attempt at capturing porpoising penguins. The 5DII is the Sony’s equal in daylight focusing. No better, no worse.
Since returning from Antarctica, however, I have used the 5DII in very low light on a number of occasions, and been rather frustrated by its autofocus ability. Shooting with a 50mm f1.4 lens in ISO 3200 1/30th at f1.4 light, I did not find the centre AF point to be ‘grabby’ enough. It works, but I have had enough out of focus frames to think some improvement is called for. To be fair, this is really, really low light. Not that long ago, we wouldn’t have bothered to shoot at all in such circumstances, but since Canon has created a camera that can produce an excellent image under these lighting conditions, it seems natural to expect it to focus without complaint.
This is another place where I found the 5DII bumping into its own performance envelope. Sensor performance now exceeds AF ability in low light. Canon R&D take note: now that it can shoot in the dark, make it focus in the dark, too please.
One difficulty I encountered was using the “AI Focus” mode in my typical focus-and-recompose work-style. After a day of critically unsharp images, I realized that the camera was often refocusing ever so slightly when I had recomposed and positioned the central sensor over a different area of the frame, which fooled the camera into thinking the subject had gone into motion. I converted to using one-shot AF with the center points for virtually everything, and had no more problems.
The outside sensors were also given a workout when shooting close-ups of Gentoo penguins at Port Lockroy (one of the few places where penguins may be approached very closely because of their choice to nest inches from the door of the British base). Using a darkish 100-400mm lens at its longest reach, these non-cross type sensors worked just fine.
Focus-tracking on swimming penguins was a greater challenge, though still with a decent success-rate. While I have not had the chance to really put the 5DII through its paces in a fast-moving sports situation, it does not appear that any of the focusing gremlins which have plagued the 1DIII found their way into this camera.
Can the AF otherwise be improved? Of course. Where’s the eye-control? That technology was around in the early 1990s on the EOS A2E and worked respectably even in its first generation. This was a feature that should have been pursued and perfected, because changing focus-points on-the-fly is simply impossible. While we’re at it, how about some focus sensors on the edge of the frame, and not just the edges of the middle? Four points in the 1/3 corner intersections would be nice, and oh-so compositionally convenient.
Of course, only real photographers would know this, and since Canon is still putting a “Direct Print” button on a 21MP full-frame camera, I can only assume that real photographers are in short supply in the design and specification departments. This is hardly surprising, since everybody knows that photographers are hard to come by in Japan.
Sarcasm aside, there is still something for the development teams to work on now that image resolution has hit its useable peak.
This what it’s all about, isn’t it? The short answer to the IQ question is that the 5DII produces images exactly the equal of the 1DSIII. The even more irritatingly unhelpful answer is that it produces images which are, for the most part, functionally indistinguishable from the A900 and D3x. Sorry folks, there’s no holy grail. (The one caveat to this is that I have not tested the D3x in 14-bit mode, and am deeply curious as to whether it makes a difference or not. Since the difference would probably show up most noticeably with really heavy file manipulation, I don’t find any of the passing dismissals of Nikon’s 14-bit capacity to be terribly informative).
But let’s break this down; what is coming out of the camera in terms of noise and resolution?
There’s lots of it. Like really a lot. How much I doubt I will ever know, since few if any of Canon’s lenses appear able to feed this sensor the detail it is capable of devouring. Specifically, I was very disappointed in my choice of taking a 100-400mm zoom instead of a fixed 400mm f5.6. Fully 85% of my keepers from this lens were shot at 400mm. At that focal length, the 100-400mm simply doesn’t have the gas. Notwithstanding its branding as an “L” lens, it’s just not up to meeting the resolution challenge offered by the 5DII. The shots I produced in 2007 with the 400mm on a 1DsII show greater enlargeability. This is a serious let-down.
I was also disappointed by a lot of the shots taken with the 24-70 f2.8L under 50mm. The only lens which performed to the resolution of the camera was the 70-200m f2.8L, which is not surprising given its stellar reputation. But even then, I can’t help but feeling that a very aggressive anti-aliasing filter is at work. While the 5DII produced beautiful files, they lack the “bite” in the fine detail that a perfectly exposed Phase One files shows. While I have not examined any D3x files, I can say that the same is true for the Sony A900 in most instances.
What does all this mean? To be blunt, Canon needs to build a lot better lenses. And a lot of them.
If you’ve been reading closely, you’ll notice that very little of the foregoing discussion on resolution had much to do with the camera. That’s because the limits of resolution performance are now dictated by lens performance and optical stability, not the camera’s sensor. My gut feeling is that I really don’t know what the 5DII can really produce in terms of raw, eye-ball lacerating resolution. If anybody wants to lend me some Leica glass, I’ll let you know.
As it is, in the real world where most photographers use zoom lenses, I would question the real value in the jump from 16-18MP to the 22-25MP range. I’m simply skeptical about how much actual quality is being gained. In studio, under strobes, with a prime lens at optimum aperture, I’m certain the 5DII will visibly outperform lesser-spec’d cameras. But in the field, files from my 1DsII mostly match its performance with any “L” zooms I own.
It’s perversely ironic that Leica this week announced cessation of their “R” line of lenses, since ne plus ultra optics have never been more relevant. The real advantage of 35mm systems lies in the vast availability of lenses – especially zooms. Lens quality is now, by far, the most limiting function in the photographic equation. To truly wring the last drops of quality out of cameras in this class, one really has to use super high-quality glass, coupled to some form of optical stabilization. While what one can achieve with the better zooms is very, very good, we are at a point in camera evolution where the last 5% really shows. This is true of some simply stunning images I have seen taken with the superlative 135mm f1.8 Zeiss on Michael’s A900 and the 85mm f1.2L on the 5D2
It is hard to understand how the obvious synergy of Leica’s optical excellence and Canon’s ability to build pixel-starved sensors has gone ignored for so long. Shear, stupid corporate pride appears to be the only plausible explanation.
To put this all in context, however, let’s remember that the baseline of quality from which the conversation now begins is already far higher than most photographers will ever need. In commercial print reproduction, a Canon G10 now pumps out more than enough pixels for a double-truck layout. Printed on inkjet or photographic paper, images from just about any higher-end DSLR will look great up to 11x14. It is only in the rarified air of really, really big prints, done with really, really good post-processing, on really, really well profiled systems that the differences we are speaking of will be visible.
To some extent, we are all quibbling about the price of champagne in the promised land, and we should remember that.
If resolution is becoming a bit moot, then the real action has moved to ISO. A miniscule amount of blur will erase acres of effective resolution, so the ability to shoot fast in low light really is now the real frontier in camera development. Here, the 5DII really shines. Going back to files from my 1DsII, I was a little shocked to realize that noise becomes a visible issue at speeds as low as ISO 500. To be fair, this is noise that can be dealt with in post-processing at the cost of a little resolution. Nevertheless, it’s a real issue in landscape photography where micro detail is the difference between good and great.
I pushed the 5D2II hard, shooting as if ISO didn’t matter. I shot casually at ISO 640, 1000, and up. Working with long lenses for a moving platform, this secured me many images I would not have otherwise gotten. But ISO does still matter. While shots up to 1600 are remarkably ‘clean’, there are traces of noise. Moreover, my subjective impression is that higher ISO shots are less malleable, ‘fracturing’ into digital artifacts and solarization after less manipulation than shots at lowers ISOs. This just stands to reason, since ISOs over the sensor’s base sensitivity are obviously achieved with less absolute true data.
This is not a criticism of the camera’s performance. The 5DII produces cleaner high-ISO files than we’ve ever seen before, and truly moves the bar on what can be thought of as usable light. Just don’t expect miracles. Or at least not major miracles.
As a former photojournalist, I was curious what the camera could do at a kids’ swim meet held in the dungeonesque community pool. When correctly exposed, shots at ISO 6400 were remarkably good – comparable to what I would have expected from 1600 ASA colour film a decade ago. I cannot think of a situation in which the 5DII could not produce a serviceable newspaper photo if armed with fast lenses. The D3/D700 combination still yields a slightly cleaner image in the ISO-stratosphere, but the 5DII is a close second, with an appreciable resolution advantage.
That said, the limits are still there. One twilight in the Gerlache Strait, we encountered a pod of playful whales. The lighting conditions were a veritable ISO torture-test: grey, flat, low, blue light yielding 1/2000th @ f5.6, ISO 2500 (from experience, the minimum shutter speed that will truly freeze a whale mid-breach with a 400mm lens).
No problem, right? Wrong. Thanks to an unimpressive histogram (read a one to two stop under-exposure using the “expose to the right’ principle), the resulting image is noisy as hell when viewed at 100%. I processed it with significant colour-noise suppression in Lightroom, and then completely desaturated both purple and magenta. This controlled the colour-noise relatively well, but left a great detail of luminance noise. But how does it look in print? Awful. Experiments with various noise-reduction plug-ins yielded equally mediocre results. In Black and white, the image has some charm, but let’s be honest, we’ve hit a limit.
Intangible File Quality
Actual photographers, as distinct from the measurbators, often speak of the ‘feel’ of the files produced by a given camera. This subjective catch-all comprises colour palette, tonal reproduction and the like. At its base ISO, the 5DII produces a very rich file. The studio shoots I have done with it have yielded extremely pleasing results. Colour is accurate and the files have reasonable depth to handle manipulation. I have trouble imagining any studio shooter being unhappy with the 5DII’s output. While I never felt the 5D was a really serious pro camera, the 5DII unquestionably is.
That said, I personally do not feel that any DSLR matches MF captures. There is a unique ‘depth’ to high-bit MF files. In Antarctica, I had the chance to shoot on numerous occasions with Phase-equipped cameras. The bottom line is that, while I struggled to get critically sharp images (a combination of needing higher shutter speed to overcome vibrations, mediocre AF systems and crappy focusing screens), the files produced by the Phase backs were palpably richer in tonality. I could stretch and distinguish detail in every area of the photograph beyond what a DSLR would allow. This makes an immediate, if hard to quantify, difference to the files. The catch is that the Phase back gets visibly noisy at ISO 400, and cannot be mated to an image-stabilized camera. Consequently, these amazing imagers were mostly retired to benches when the best light of evening and dawn rolled around.
Bottom line: in the inevitable world of compromises that is photography, the 5DII lets photographers shoot better in lower light than ever before. The limits are there, but they’ve been given another good, hard shove towards the corner.
Odds and ends
I shot just under 4,000 frames on the 5DII during the two weeks in Antarctica. While sea-sickness medication and twenty-one hour shooting days are not conducive to mental acuity, I remember charging one of the 5DII’s batteries on four occasions. Neither ever hit ‘empty’. Whatever the math works out to be, the 5DII was getting well over 500 frames per battery. This was with heavy use of IS lenses. Battery sufficiency is just not an issue.
By comparison, I shot slightly fewer frames on the 1DsIII over the same period of time, but charged only one of the two batteries I had for it on a single occasion, translating to a yield of closer to 1,000 frames on the ‘pro’ camera’s much larger cells. On a simple battery-mass analysis, both are performing admirably.
The auto white balance works almost flawlessly, but that’s hardly news. If you are in a colour-critical application, you will shoot a grey-card. If not, the camera will take good care of you.
The Inevitable Comparisons
Michael has said it best: the margins between the top-end cameras are just too close to call nowadays. Which camera you get is largely a function of personal taste, need, and previous investment in a system of lenses. Switching between Canon, Sony and Nikon’s top-end offerings will improve nothing but your relationship with your camera dealer.
I like the 5DII a lot, but I don’t love it. It’s still a plastic computer hooked up to a system of solidly B+ lenses. Put this sensor in a Contax RTS III, and I’d be in heaven. But we photographers are never satisfied….and that’s the fun of it!