It says a lot about the Canon EOS 5D that, years after its debut, the release of a successor does not necessarily compel 5D users to upgrade, even though the new model packs 21 megapixels, improved high ISO performance, and burst speed, high-def video, and more. The fact is, the original 5D is a great camera, and many users are coming to realize that they may have all the pixels they need for the type of output they do. However, the 5D Mark II’s improved noise performance and interface changes are good enough that, depending on what and how you shoot, the cost of an upgrade might be worth it. And, while you don’t need 21 megapixels for every shot, the ability to print large, and crop tight, can be welcome in many situations. In Part 1, I looked at the interface and physical changes between the 5D and the 5D Mark II. Now it’s time to discuss image quality.
Before jumping headlong into Mark II image quality, I’d like to mention one additional change that I didn’t cover in part 1. Canon claims that the Mark II has an "improved" autofocus mechanism. Before the camera was released, many 5D owners expressed concern that Canon had not installed a better autofocus system in the Mark II. While the 5D’s autofocus is quite capable, it definitely falls down in low light, and doesn’t perform as well as competing systems from Nikon.
In my experience, though, Canon is correct: the autofocus is definitely improved in the Mark II, especially in low light. It’s not a new, dramatic, magical difference – you won’t find yourself able to focus in absolute darkness – but you should find yourself less frustrated than you were with the 5D. While not a revolutionary change, the Mark II’s autofocus is definitely improved, and this is a change that makes the upgrade more compelling.
Why there are few image examples in this story
Image quality is, of course, a very subjective topic. Most online camera reviews focus on a few areas when assessing camera quality: sharpness, noise, and overall color accuracy.
It is, of course, possible to shoot color charts and measure results with specialized gear to get a measure of the accuracy of a camera, but accuracy is just one consideration when evaluating an image, and some subjects are not well-served by accuracy. Trying to deliver color testing samples via the web is an exercise in frustration due to the lack of any ability to calibrate my system with the reader’s. Sure, you can tell relative changes between two images on the same monitor, but if that monitor is wildly out-of-whack, this will be only an approximation of overall quality.
Sharpness and noise can also be measured using one technique or another, but most reviewers tend to evaluate these qualities by examining 100%, pixel-for-pixel crops of test images. Personally, I find such examinations to be a meaningless exercise, and I think it’s a shame that these examples have become so popular, and are taken so seriously.
If you’re buying a $3000 camera, my assumption is that it’s because you’re serious about photography, and that your goal is ultimately to create photographic prints – either fine art prints, or prints in a publication. In either of these instances, the only noise that matters is the noise that appears in your final print. When you look at the individual pixels in an image from a 12 or 21 megapixel camera, you’re seeing pixels that are most likely not visible in a final print. The physical size of an individual pixel is TINY, so evaluating noise or edge detail by examining individual pixels is roughly akin to judging print quality in a magazine by examining the halftone patterns used to create specific colors.
For my 5D/5D Mark II comparisons I judged image quality from prints. Because there’s no accurate way to make these prints available, I’m afraid I can offer no hard data to back up my conclusions. Consider this evaluation as a single data point in your buying decision. Check out other reviews and other opinions, and see if you can find any commonality.
Lenses and the 5D Mark II
While a sensor with a higher pixel count lets you print larger, and crop smaller, it is also, potentially, less forgiving of lens aberrations and faults. To fit more pixels onto a sensor of a given size, those pixels must be made smaller, and the micro-lenses attached to each pixel must be redesigned. These changes can, theoretically, mar the performance of a lens.
If you’ve been shooting with the 5D for a while, then you might have built up a lens collection. The good news is that you’ll probably see no trade-off in quality when you upgrade to the Mark II. I shot both the 5D and the 5D Mark II with the Canon 50mm 1.2, the Canon 24-105L, the Canon 16-35L II, and the Tamron 28-75 2.8. As expected, the Mark II showed better detail, thanks to its higher pixel count, but overall sharpness, vignetting, distortion, and image quality was the same with each lens on both cameras.
As with any camera, putting a better-quality lens on the Mark II will give you a better image. If you’re shooting with a marginal lens, you’ll see marginal quality. But you don’t need to worry about the Mark II "degrading" your lens in any way.
Canon digital SLRs have been lauded for their extremely low noise since the original EOS D30, and with each successive generation they have further improved the camera’s ability to shoot at high ISOs, in extremely low light, and yield an image with extremely low noise.
When you take a picture, one of the first things that happens inside your digital camera is that the data is read off of the image sensor and then amplified. When you increase the ISO setting on your camera, the amplification of the data coming off of the sensor is increased. While this serves to make the sensor more sensitive, it also makes any noise in the image more pronounced.
The amount of noise is a function of several factors, the most significant being pixel size, and image processing. Typically, when you reduce pixel size, which you have to do to increase the pixel count of an image sensor, you run the risk of increasing noise. Consequently, when the Mark II was announced, many people worried that its higher pixel count would compromise the low noise levels of the 5D.
While the Mark II does pack a tremendous number of pixels, its larger sensor size means that the pixel size and density is actually no greater than on Canon’s EOS 50D. The good news for current 5D owners is that the Mark II not only maintains its predecessor’s excellent low noise levels, it actually improves upon them at higher ISO’s.
There’s not a significant improvement over the 5D up to ISO 1600, and this is largely because the 5D was already so good at these levels. It’s at ISO’s higher than 1600 that the Mark II amazes.
On the original 5D, you could activate "ISO Expansion" which allowed you to set the ISO to "H", which was the equivalent of ISO 3200. While this setting wasn’t bad, I found it a little too chunky and noisy for general use. On the Mark II, though, ISO 3200 is surprisingly clean, and definitely very usable. Also, you can access ISO 3200 directly, as well as ISO 6400. At 6400, things start to break down a little bit, with 6400 on the Mark II being slightly better than 3200 was on the original 5D. While not overburdened with luminance noise, chrominance noise – colored splotchy noise – is fairly obvious at ISO 6400. However, much of it gets sampled away or hidden when printing, so you’ll want to try some prints before you decide on the viability of shooting at this higher speed.
When you activate ISO Expansion on the Mark II, two additional ISO settings will become available beyond 6400. H1 is equivalent to ISO 12,800, while H2 gets you ISO 25,600. As with ISO 3200 on the original 5D, Canon has kind of "hidden" these settings because they’re pretty grungy, though still very usable, depending on your aesthetic and imaging needs.
ISO Expansion also opens up the same L option that you had on the 5D. This is the equivalent of ISO 50. Given how clean the camera’s ISO 100 is, you’ll probably rarely find yourself using this option.
There is one important caveat to all this noise stuff, which is that Canon is cheating a little bit. When you shoot the 5D and the 5D Mark II side-by-side, with the same settings, at the same ISO, you’ll see that the Mark II consistently shoots roughly one half to two-thirds of a stop darker. In other words, the ISO improvement isn’t quite as big as it would seem, because Canon is fudging their ISO ratings somewhat.
For most users this won’t present any issues of any kind. However, if you’re planning on keeping your current 5D and using it as a second body – say, for event shooting – then you’ll want to do some experimenting with ISO settings and exposure compensation between the two cameras. You might find that you need to set the Mark II’s ISO a few fractions of a stop higher than the 5D, to ensure that images from both cameras have equivalent exposures.
All-in-all, the Mark II continues the advancements that have been progressing in Canon digital SLRs for the last eight years. The Mark II yields better high ISO performance than any camera Canon has previously released. The ability to shoot at ISO 3200 and 6400 is fantastic for low-light shooters of all kinds, and this feature alone might be reason enough for you to upgrade, depending on the type of shooting you do.
Raw and sRaw
While many people get excited about the buckets of pixels that the 5D Mark II can pour into an image, other people see it as necessitating more storage and a faster computer. Because 21 megapixels is often FAR more image data than you need, the Mark II can also shoot in two additional raw formats, sRaw 1 and sRaw 2. ("S" is for small, by the way.)
The normal raw mode yields a 21 megapixel image, with dimensions of 5616 x 3744 pixels. On average, one of these files will consume 25.8 megabytes, giving you space for approximately 72 images on a 2 gb card.
sRaw 1 mode yields a ten megapixel image, with dimensions of 3861 x 2574 pixels. These files will take up 15 megabytes of storage, or 120 images per 2 gb card.
Finally, sRaw 2 mode will give you a 5.2 megapixel image, with 2784 x 1856 pixels, and will consume 10.8 megabytes per image, giving you 170 images per card.
Creating a smaller raw file is not an easy thing. Raw files don’t actually contain full color image data. Instead, the raw conversion software that you use calculates the color of each pixel through a complex interpolation process. Because the color of each pixel is calculated based on the values of its neighbors, you can’t simply throw out pixels to reduce the pixel dimensions of the file.
The sRaw files have all of the advantages of normal raw files – highlight recovery, adjustable white balance, full 14-bit color depth – but at half or one quarter of the resolution. This means you can save a lot of storage space, if you don’t need a full-res file.
Many people have speculated that sRaw files should exhibit less noise than full raw files, through some kind of downsampling or oversampling process. In examining results, there’s no apparent difference in noise, and Canon’s own Chuck Westfall states explicitly, here, that there is no noise difference.
In my tests, though, there is a noticable color shift when you switch from raw to sRaw. sRaw images are a little warmer, with a very slight magenta cast, while normal raw files have more of a greenish cast.
The upper image was shot as a normal raw image, the lower was shot as sRaw 1. As you can see, the color shift isn’t huge, but note the gray wall in the background. That’s the easiest place to see the shift.
I processed the images in both Photoshop Camera Raw and Canon’s DPP, and found similar problems in both. sRaw 2 has an additional difference in that it crops your image a tiny bit tighter than either of the other two modes.
For event shooters, this color shift could be a hassle, if your goal is to shoot some parts of an event as full raw files, and other parts as sRaw.
For those who are wondering if sRaw images are sharper than regular raws, I found little discernible difference in sharpness. Sampling a full raw image down to sRaw size yielded a sharper file than shooting in sRaw directly.
Overall Image Quality
As with its predecessor, the 5D Mark II yields great images. With its new DIGIC IV processor, the Mark II outputs 14-bit raw files. While this won’t let you see any more color or dynamic range than you saw on the 5D, it does provide you with more editing latitude. You’ll be able to push edits farther before you see noticeable posterizing or tone breaks.
Early 5D Mark II users reported problems with a "black dot problem" – black dots that could appear next to bright lights, especially when shooting in darker situations. I never encountered this problem, but Canon has already released a firmware update that fixes the issue, as well as a vertical banding problem that some users reported when shooting in sRaw. When you get a new camera, you’ll want to ensure it’s using the latest firmware (1.0.7, at the time of this writing). If it’s not, then you’ll need to download the newest, here.
Borrowing another feature from the higher end of the EOS line, the Mark II offers the same micro-focus adjustment feature that is provided in the 1DS Mark III and 1D Mark III. This allows you to calibrate the focus of your specific lenses to your Mark II body. The process is fairly simple, and is implemented through a custom function. If you’ve invested in expensive lenses, being able to ensure that they’re focusing as well as possible is a nice feature.
All-in-all, the ISO improvements, higher bit depth, and increased pixel count make for images that are simply better than the 5D’s when shooting in situations, or outputting to situations, that require better ISO, bigger edits, or more pixels.
Also new with the 5D Mark II is the ability to shoot video in either standard or high definition. As discussed in part 1, there are fairly pronounced usability differences when you choose to shoot video with an SLR rather than a video camera. There are also some quality issues that you’ll want to consider.
Because there’s really no way of taking any manual control of the camera’s exposure settings, experienced videographers might find themselves frustrated. While shooting, you can turn the rear control wheel to adjust exposure compensation, but this is the extent of the exposure control that the camera provides. Unless you have a lens with a manual focus ring, (which means an older Nikon lens and an adapter) the only way you’ll be able to control aperture or shutter speed is by tricking the camera.
With Canon autofocus lenses, the 5D can only shoot at four different apertures, f22, f16, f5.6, and the widest aperture of your lens. By holding your hand in front of the camera, you can force it to change ISO and aperture, and then use AE Lock to lock that exposure. You can then use the rear control wheel to adjust up or down. By fiddling this way, you can force the camera to choose, for example, your widest aperture for shallow depth of field.
These techniques can work, but they’re not especially streamlined or quick to implement. Because there’s no easy way to lock down aperture or shutter speed, there’s no guarantee that your depth of field won’t change unexpectedly, or that you won’t have a fast, stuttery shutter speed.
Finally, there are times when the 5D exhibits a slight rolling shutter. That is, when you pan, the top part of the frame pans faster than the bottom.
If you’re just shooting podcasts or simple documentary bits, these problems are no big deal, and the camera truly does record very good video. The inclusion of an external mic jacks was a great decision on Canon’s part, as it means you CAN control the quality of the audio you’re recording. If you’re looking to shoot more serious work, with precise depth of field and/or shutter speed control, know that it can be done, but it’s a hassle.
The advantage of shooting video with the 5D, of course, is the camera’s exceptional low light performance, and the ability to shoot with the high quality lenses that you already have. The ability to use fast lenses means you can shoot with extremely shallow depth of field, while telephoto, tilt/shift, macro, and other specialty lenses give you additional options that are simply impossible on conventional video cameras.
The final decision
Unfortunately, I can’t say for sure whether the upgrade is a good idea for you. Everyone has different needs, depending on the type of shooting they do. Hopefully, the points that I’ve raised here will help you understand what questions you need to answer before making your final decision.
I’m very happy with my upgrade, as I love the better low light performance and faster burst speed for event shooting. For landscapes you can never have too many pixels, especially if you like to print big, and the ability to save custom shooting modes means I can easily switch between lots of commonly used configurations.
Fortunately, you won’t really give anything up by upgrading, except a pile of money, so you don’t have to worry about trade-offs. If the new features sound like they’re worth the money, then you should probably upgrade. The good news is that, even if you decide not to, the original 5D is still a great camera. In the end, you can take great pictures with either.