For the first time this decade, there’s a new camera in my main photography bag. It’s not that I’m afraid of new technology. Actually, I typically embrace new tools quickly if they make a meaningful difference in my work.
And that’s been the issue. Since the Canon 1Ds Mark III was released in late 2007 there hasn’t been a compelling reason for me to upgrade my primary landscape camera. There have been video features and improved low-light performance, but nothing that was a dramatic leap that would change the way I work. For nearly 8 years, resolution from Canon’s top cameras has been virtually unchanged.
The new Canon 5Ds R finally delivers a huge leap in image quality. It’s a 50-megapixel camera (my 1Ds was 21). The narrow edge of its files contains more detail than the long edge of the files from my old camera. Combined with video and improved lower-light capabilities, it was finally time for me to upgrade.
This post describes my initial experience with the Canon 5Ds R as it relates to the work I produce and how I produce it. I cannot compare it to the Canon 5D Mark III as that is not a camera I have used on a regular basis. I also cannot compare it to Nikon or Sony models as I haven’t been interested in replacing all my lenses or dealing with adapters. I also have not shot test charts, although others have. This is about my real-world experience with a production model of the new camera.
Who needs 50 megapixels?
One of the most common refrains in Internet forums is, “Who needs 50 megapixels?” I cannot answer for the whole Internet, but for me, personally, the answer is, “I do.” While websites and other low resolution uses are making up a greater percentage of my overall image licenses, large prints generate more of my income. Having more pixels means I have more options.
While I have been able to turn even some 11 megapixel files into 11-foot murals, they have been very special cases that involved unique images, very sharp files and slow enlargement software. Having more resolution gives me more options for every image, which is important when I rely on my image library for income and a request for a giant usage may not come in until years later. I have never once thought, “I wish this image file contained less detail.” I can always reduce files after the fact, but you can never restore details your camera wasn’t able to capture in the first place.
Is the detail noticeable?
The nature of my work — photographing active wildlife and landscapes in changing light — doesn’t make it easy to do side-by-side comparisons. Subjectively, as I open the Canon 5Ds R files on the 27-inch monitor in my office, I am blown away by the detail that’s in them.
I can easily read the name on a ferry that’s halfway across Puget Sound. I can count the number of petals on a small wildflower that takes up less than 1 percent of the space in an image of a waterfall.
I was able to do one comparison between the 1Ds Mark III and the 5Ds R, although the images were captured several minutes apart as the sun was rising. While the lighting is different, this comparison does give you an idea of the difference between a 21 and a 50 megapixel camera.
Both of these examples are 100% crops. Because of the increase in resolution, the 5Ds R image appears to be zoomed in. Both images are aligned on the top-left corner.
1Ds Mark III:
If you look close, you can see some very narrow layers of debris in the glacier in the 5Ds R image that simply aren’t visible in the 1Ds Mark III image.
What about moiré?
The Canon 5Ds comes in two versions because of a quirk of digital photography. A camera may have 50 million pixels, but each one sees only one color — green, red or blue. It’s not a Canon thing; nearly every digital camera has this limitation. The camera’s processor tries to determine what the actual color is for each pixel by using an algorithm that considers the color of the neighboring pixels. Camera makers have traditionally aided that processing by putting a blurring filter over the sensor so that each pixel also gets a bit of its neighbor’s light too.
The version of the camera that I have, the 5Ds R doesn’t have the blurring filter — or, more accurately, has the filter but also has a second filter that cancels it out. The idea is that images would be a little bit sharper, since there’s no blurring going on. The concern, however, is that without the filter to soften the light a bit, in images with very fine patterns, the algorithm may be fooled and produce colors and other artifacts that aren’t really there.
These false colors are most likely to show up in pictures of fabric, which consist exclusively of very fine repeating lines, and significantly less likely in nature, which has no straight lines. I figured bird feathers are the closest things in nature to fabric and I photographed a number of birds in an attempt to produce moiré. No luck — and that’s a good thing.
In fact, after two months of heavy use, I noticed only one image that exhibited any sort of a flaw. It was a case of aliasing, where what is supposed to be a straight line can trigger some stray pixels.
It occurred a picture of a ferry boat taken at great distance. Part of the railing appears jagged and pixelated in the image in a way that I’ve never seen before from a digital camera. You can see it here through a very extreme enlargement. Even after seeing the flaw here, it’s not something I’m the least bit worried about. I couldn’t even find it in the frame before or after in the series. And it’s so small, you would never notice it in print.
If you’re willing to accept that small risk, the 5Ds R, in certain cases, does deliver more resolution than the 5Ds. I don’t have the 5Ds so I’m not able to run comparison tests, but Bryan Carnathan at The Digital Picture did. Take note of the third outdoor scene in his sharpness test. Siding that is clearly visible on the building in the 5Ds R image is nearly blurred away in the image from the 5Ds.
The camera that brings out the worst in lenses
There has been tremendous debate online over whether any given lens or certain techniques are up for the leap in resolution that this camera offers. Canon has gone so far as to publish a list of lenses that are acceptable for this camera — the Canon EF 600 f/4 IS, one of the sharpest lenses I’ve used, didn’t even make the cut. At a briefing before I got my camera, a Canon technical advisor talked about how the camera was really intended for tripod-use and that handheld images may appear subpar — the camera is so sharp you would easily see any bad technique.
The debate has been fierce. One camp says all that talk is just a bunch of marketing hype. The other says it’s only natural that the more detail you capture, the more any flaws would be magnified.
After using the camera for more than two months, I fall somewhere in the middle of the debate. I believe the list of recommended lenses is far shorter than it should be. My one-generation-old 600mm performed flawlessly on the new camera body and delivered sharper images than I’d seen from other my 1Ds Mark III or 1D Mark IV. That lens is not on Canon’s 5Ds approved list.
But to a degree I also believe there are good intentions behind Canon’s advice. If you really zoom into the files and look at them at a pixel level, you will see flaws that won’t be as noticeable on a camera that offers less resolution.
The Canon TS-E 24mm tilt-shift lens is one example. It was introduced in 1991 and compared to more recent lenses, you can easily see how much detail it's lacking. This animation compares that lens with the relatively new EF 24-70 Mark II. The image from the tilt-shift lens was captured with a small amount of shift; and the 24-70 zoom lens was set to 27mm, so this isn't a perfect apples-to-apples comparison. Both, however, were set to an aperture of f/7. Still, the high-resolution sensor shows an extreme difference in image quality between the two lenses.
And here's another example. I know that using a very tiny aperture, like f/22, impairs image quality. Such tiny apertures cause the light that passes through to bend, blurring the image. While not ideal, sometimes you need to use these apertures to maximize depth of field. It’s a tradeoff, and sometimes it’s better to give up a little sharpness to have more of the image in focus.
I shot a landscape image with the Canon 5Ds R at f/22. I had some flowers in the foreground that I wanted in focus and a mountain range in the back that needed to be in focus too. When I zoom all the way into the image, I was struck by how much detail was lost due to diffraction. I was stunned at how soft the image was. I had always known about diffraction, but had never seen its blurring effects so clearly.
But then I reduced the file down to the same dimensions as the files from my old 1Ds Mark III. And the result looked a lot like what I would have expected from the old camera. The Canon 5Ds R wasn’t worse in any way.
I think what Canon means is that if you want to get every last bit of detail out of your files, you do have to step up your game. If you are shooting without a tripod, you may want to boost the shutter speed a bit more than you have in the past.
In my case, instead of using f/22, I’m thinking about investing in more tilt-shift lenses, which allow you to position where the focus depth occurs, allowing you to shoot with wider apertures that produce less diffraction. Focus stacking — shooting several images at different focus settings and combining the sharp parts from each image later in Photoshop — is another option.
If a lens or particular technique delivered acceptable results for you with your old camera, the results will be at least as good, and possibly a bit better, with the new Canon 5Ds. But don’t expect marginal lenses or technique to produce images that are twice as good with the new camera.
Mirror lock-up matters
While the Canon 5Ds R looks little changed from the earlier Canon 5D Mark III, there are a number of internal changes that are supposed to help deliver maximum results from the high-resolution sensor.
One change is that the tripod socket has been redesigned with new material to dampen vibrations. The mechanism that moves the mirror before and after the exposure has had an even more radical redesign to make its movements less forceful. You can hear the difference. A normal exposure with my Canon 5Ds R is much quieter than a “silent mode” shot with my previous Canon camera.
Fewer vibrations should lead to sharper images and Canon has also made the mirror lock-up mode easier and more flexible to use. When mirror lock-up is on, taking a picture becomes a two-step process. First, the mirror is moved out of place. Once vibrations from that process subside, then the picture is taken.
With older Canon cameras, you either had to use a cable release and press the button twice — once to move the mirror, once to expose — or put the camera in two-second delay, which automated that process. With the delay, the mirror immediately moves out of place when you press the shutter button and then two seconds later, after the vibrations have subsided, the camera takes the picture.
With the Canon 5Ds, you have a number of new delay options: 1/8, 1/4, 1/2, 1 or 2 seconds.
Given what appears to be a dramatic improvement to the dampening of the mirror, I ran a test to see how much of a delay, if any, you need to use.
I used a tripod, cable release, 400mm lens with image stabilization off and took an image with each of the delay settings. Each image was taken at a shutter speed of 1/20, which is in middle of the range that typically benefits most from mirror lock up. Dramatically faster shutter speeds are usually quicker than the vibration, so the image is still during the exposure itself. With dramatically slower shutter speeds, the vibration occurs only during a small fraction of the overall exposure and thus is largely invisible in the final image.
My results showed that even though the new mirror mechanism is dramatically quieter, it still triggers pronounced vibrations that can ruin your images. Those vibrations, though, subside very quickly. With a delay of just 1/2 second, I was getting tack-sharp images. In my test, there was no additional benefit to using even 1 or 2 seconds of delay.
Keep in mind, however, that in this delay test, I triggered the camera with a cable release. The mirror lock up delay settings are used more commonly with the shutter button on the camera itself. Touching the camera causes a whole different series of vibrations and you probably still want to allow for the full 2 seconds of delay.
If you are using a cable release, though, you don’t have to wait very long for the internal vibrations to subside. The new design appears to work very well.
One area where Canon has really fallen behind over the past several years is with dynamic range. The past few versions of Nikon and Sony cameras have captured much more detail with less noise in the shadows than Canon’s models have. I personally know several professional nature photographers who used Canon cameras for much of their careers only to switch brands in the last year or two to get a camera that was capable of capturing a much wider range of light.
With this new release, Canon still isn’t marketing the 5Ds as a high dynamic range performer. Official statements have described its range as similar to the 5D Mark III, a camera that’s two years old. Still, even the 5D Mark III was one generation newer than the camera I had been using, so I was interested in seeing how much improvement Canon had made.
There is noticeably more detail in the shadows of the 5Ds R than there is in the Canon 1Ds Mark III. But these shadows also appear noisier than what I’ve seen colleagues capture with Nikon and Sony cameras.
I don’t plan to make a habit of this, but I underexposed a few landscape frames to retain detail in the bright moon. Processing the RAW images later, I was able to significantly boost the shadows to make a pleasing final image, pushing the shadows more than I have been able to in the past. Especially when using the camera at its lowest ISO settings, I’m finding at least an extra two-stops of usable shadow detail.
In this image, I boosted the shadows by 100 in Adobe Camera Raw. Looking at the image at the pixel level, I do see a fair amount of grain in the deep shadows, but it’s also a fine grain that can be removed easily with software. There also doesn’t appear to be any banding and there is relatively little blotchy color noise, both of which are more difficult to remove. And in the original file, everything below the mountain is nearly black.
You’ll get the greatest potential for recovering shadows in images that were taken at ISO 100. Even at ISO 800, if you push shadows in any kind of extreme manner, you will notice a lack of detail and some color noise in addition to the pattern noise.
It’s a step in the right direction. Without using stacked exposures or graduated neutral density filters, I would not have been able to create this image with my previous camera. But if you’ve been lusting after the miracles that Nikon and Sony photographers have been pulling off with the shadow recovery capabilities of their cameras, you will still find the 5Ds lacking.
This animation gives you an idea of how much you can recover from the shadows and highlights with the Canon 5Ds R. The darker image is a straight conversion in Adobe Camera Raw; the brighter image was converted with a +1 exposure, -50 highlights, +100 shadows and +100 blacks.
And here is a 100 percent crop from the extreme shadow recovery version. This is from the bottom left corner of the image, a part that looks nearly black straight out of the camera.
It's a bit noisy, sure, but keep in mind this was the result of maximizing all of the shadow recovery options in Adobe Camera Raw and adding an additional stop of exposure on top of it. This is much more than I would ever want to push an image — my final version is pushed a bit, but much closer to the original high-contrast version — but especially when shooting at ISO 100, there's a fair amount of detail in very deep shadows with this camera.
Low-light sensitivity is another area where Canon has been perceived to be lagging over the past few years. Canon isn’t even trying to compete in this area with the 5Ds; it has actually reduced the maximum ISO setting available on this camera. The 5Ds tops out at ISO 6,400; the 5D Mark III goes to 102,800.
A dramatic increase in resolution means each pixel on the sensor has to be smaller. Smaller pixels are more prone to noise. But even the Canon 7D Mark II, a relatively new camera that has pixels that are roughly the same size as the 5Ds, has a top ISO setting of 51,200 — a setting that’s 8 times as sensitive as the 5Ds’s top setting.
In my experience, the low light performance from the 5Ds isn’t likely to impress anyone who uses a Nikon or Sony camera. But it is an improvement over my previous Canon camera and it actually does a respectable job in most typical situations.
In RAW images — an important distinction that I’ll get to in the next section — there is a visible grain at settings as low as ISO 800. If you look close, you can even see it at ISO 400. It’s basically the same performance at the individual pixel level that I had gotten from my 7-year-old camera.
At ISO 1600, there’s even more grain, but this is actually where the camera started to impress me a bit. The noise is a luminance grain with limited color blotches; the image is still very rich in detail. Even at ISO 3200, there is still a lot of detail in RAW files.
This type of noise is very easy to reduce with software tools. This image is an ISO 3200 RAW image that was processed with Imagenomic Noiseware. A 100% crop of the final image still shows some grain in the out-of-focus sections, but I think the result is a very pleasing image that’s fine for publication or even large prints.
In practice, this is a camera that I will predominantly use at ISO 100 where there is no visible noise. But I’m also not going to be afraid to boost the ISO when the light is low and I need faster shutter speeds.
JPEG vs. RAW
Perhaps because the perception that Canon cameras produce noisier images than rivals, the noise reduction used in the JPEGs produced in the camera is very, very aggressive. You won’t see a lot of noise in JPEGs produced at even high ISO settings; it — and lots of detail — have been smeared away. Even the low noise reduction setting is strong.
ISO 1600 JPEG:
Processed ISO 1600 RAW:
At low ISOs, however, the JPEGs are simply stunning. Set to the landscape profile, the colors are absolutely vibrant. There is also a beautiful contrast in the middle tones that really makes the images pop. Straight out of camera, the images remind me of what I used to get from slide film.
It takes a surprising amount of effort with RAW converters to achieve a similar look. Adobe Camera Raw and Lightroom are usually my first choice, but I’ve been really disappointed with their output, especially on 5Ds R images that contain a fair amount of dynamic range. While the in-camera JPEGs contain detail in the shadows, hold the color in the highlights and still deliver nice contrast, by comparison, RAW images converted with the Adobe products look flat with bleached highlights. With images containing less dynamic range, the Adobe products are fine, however, especially when setting Clarity to 15 or so.
I’m now using Capture One more often. In the past, I had used it only for more challenging images. Its “natural punch” does a better job of delivering the contrast I am looking for, but there are still occasions where I prefer the in-camera JPEGs to images I’ve spent a fair amount of time processing. This is the first time I’ve left a camera in the RAW+JPEG shooting mode.
I initially hated the autofocus system in the Canon 5Ds R, but have since grown to love it. The 5Ds has 61 autofocus points, 41 of which are cross-type sensors. Even though it’s marketed as a studio/landscape camera, it has a more advanced autofocus system than I have in my Canon 1D Mark IV, a 6-year-old camera that is designed for photographing wildlife.
My issue with it is consistency. When it works, it’s the best autofocus system I’ve used. When it doesn’t, it’s absolutely frustrating, because it tends to struggle in what should be very easy situations. That’s what makes it truly puzzling: The more difficult the situation, the better it performs.
Since this defies all logic, I held off on posting this review until I could figure out what was wrong. For a little over a month, I have used the 5Ds R for all my non-critical wildlife photography, even taking it to a zoo to be able to practice with a wider variety of animals.
I came to this conclusion: You have to be extremely careful with your use of the autofocus expansion points. The camera sometimes doesn’t use them when you think it would, and it can get distracted by them when it doesn’t really need them.
I have noticed that if the autofocus point you select can’t acquire any focus, the 5Ds R appears to immediately give up without trying any of the surrounding points, even if you have set the camera to use them. When photographing a landscape where the sun was setting behind a hillside, the autofocus point I manually selected covered a solid area of the hill. The 5Ds R wouldn’t focus at all. Perhaps that portion of the hill was too dark, but the camera didn’t appear to even try the surrounding points as several of them straddled the top of the hill where there was strong contrast.
Other times, the 5Ds R gives those expansion points priority. I was photographing patterns in the bark of a Douglas fir tree. A nuthatch landed on the tree and climbed into one of the grooves. I manually selected an autofocus point that was entirely within the bird, but didn’t have time to turn off the expansion points. Every shot was out of focus. The 5Ds R instead used the surrounding points to focus on the edges of the bark, which were closer than the bird.
This priority the camera seems to give surrounding points is a consistent problem that spoiled a number of bird-in-flight and other action images. The problem remained regardless of the tracking sensitivity setting I used.
After I switched to the minimum number of expansion points, the camera has become a stellar performer. Images are consistently in sharp focus, even when the subject is a bird that’s flying erratically.
For testing purposes, I decided to track a kingfisher flying through a marsh. It was often 50 to 100 feet from me. At times in the viewfinder, the bird was only the size of an autofocus point indicator. Even though the bird was flying in front of some cattails and behind others and at an angle toward me — a very challenging test of an autofocus system — every image was sharp.
The results are worse if the bird is similar in color to the background. The kingfisher had a bright white patch on its chest. The 5Ds R didn’t perform as well with a Cooper’s hawk, which has a brown coloration like the marsh plants that were in the background. In a burst of seven frames, my selected point missed the bird on the fourth frame. Every shot after that was focused on the background, even though the autofocus point was back on the bird by the very next frame and the autofocus setting was set the mode that’s supposed to be most forgiving of one miss.
I now have absolutely no reservations about using the Canon 5Ds R for even critical wildlife work, but I would be even happier if the camera was smarter about when it made use of autofocus expansion points.
Autofocus can be blinded by the light
The 5Ds R is less forgiving of having the sun in the frame than other Canon cameras I have used. If the edge of the sun is within about a 30-degree angle of the autofocus point you selected, I’ve found the camera will often refuse to focus when using a lens that’s longer than about 70mm.
This struggle occurs even when the sun is filtered by moderately thick clouds or smoke. The sunsets near me have been very colorful lately due to wildfires a few hundred miles away. The smoke and haze have blocked just enough of the sun so you can look directly at it even an hour before it sets. That amount of light can still overwhelm the 5Ds R autofocus system.
Other wildlife photography considerations
The Canon 5Ds and 5Ds R can shoot up to five frames per second. By today’s standards, that’s not speedy. Canon has other dSLRs — some that are cheaper — that can shoot twice as fast or even faster.
But the camera is also more capable for wildlife photography than that specification might have you believe. My first dSLR could capture only three frames per second and I managed to produce some fantastic action wildlife photography with it.
When producing RAW files, the 5Ds R can capture 14 frames before its buffer is full. That drops to 12 frames if you’re producing RAW and JPEG files simultaneously.
For my tests, I exclusively used a Lexar Pro 1066x 64 GB CompactFlash memory card, which is one of the fastest with the 5Ds R. I perhaps shot a little more deliberately than I do with my wildlife camera, but I never ran into a situation where the 5Ds R wasn’t ready to take a picture. The buffer seemed to clear very rapidly.
It is quite slow, however, when it comes to reviewing images after you’ve shot a burst. The back display can be locked up for 4 or 5 seconds. Even when the back screen isn’t available, the camera is still ready to take more pictures.
Since completing the tests, I have switched to Lexar Pro 1066x 128 GB CompactFlash cards, which are nearly as fast, but give me twice as much storage space.
If there is one area where the Canon 5Ds R is a step backward, its battery life. With my Canon 1Ds Mark III and 1D Mark IV, I had spare batteries but never used them. With the 5Ds R, I will be buying several additional spare batteries before traveling with it.
I have not let the camera run completely dry. These figures are based on the battery performance figures available in the camera’s menu system, which show the remaining capacity and the number of images taken since the battery was last charged.
If the battery figures are accurate, I am averaging between 700 and 750 images per charge. That’s moderately less than what friends of mine get with their 5D Mark IIIs, and is about half as many shots as I got with the 1Ds Mark III. Granted, the 1Ds Mark III battery is larger, but only about 25 percent larger.
I’m wondering if the display or the electronics that feed it are the culprit. I rarely review images in the field, but I do have the camera set to briefly flash the image on the screen after capture — the camera’s default setting. During a session where I was doing a lot of action photography — where only the last frame of a burst would briefly display on the screen — I had captured 827 frames and still had 32 percent battery remaining. That would put me on track for 1,240 shots on a single charge.
Despite that, I have no plans to turn off the confirmation display. I once had a memory card fail. It was a different brand of card in a different brand of camera, but that confirmation feature immediately clued me in to the fact that something was wrong. While cards can fail at any time, I do find a little comfort in seeing the image I captured briefly flash before my eyes.
If you’re wondering if there’s a difference in 50 megapixel files, there is. There is noticeably more detail. But the thing that wows me the most is the color rendition of the Canon 5Ds R. The files are often vibrant and gorgeous straight out of the camera. It’s the first digital camera I’ve used where I’ve immediately felt like I was looking at slides.
The files do contain more dynamic range and the shadows have less noise, giving you more post processing options. But here, you’re looking at an incremental improvement. It’s not game changing.
The same goes for the camera’s low-light capabilities. Images created at less than ISO 800 are smooth and noise free. I have no problem using ISO 800 and ISO 1600 images with a bit of noise reduction on the RAW files. ISO 3200 can be used as well, depending on the subject matter. In all cases, I found the in-camera noise reduction too aggressive for my tastes.
While the camera is marketed as a studio/landscape camera, it is surprisingly capable for wildlife photography. If you’re relatively deliberate and can anticipate action, the camera can easily keep up with you and deliver sharp images. As with all Canon cameras, the autofocus system needs some tweaking to your personal tastes and with this camera, the expansion points can really hurt you more often than they help.
In addition to spare memory cards, you’ll need spare batteries, even if you’ve never had a camera where you needed them before. Its battery performance reminds me more of the Canon D60 — a camera that came out in 2002 — than any recent model.
Things like diffraction and sloppy technique do ruin images faster than with lower resolution cameras. It’s simply a matter of having more detail. If you shrink the files down, those flaws become markedly less visible. But if you’re shrinking the files down, why would you want this camera in the first place? You’ll get the most from it if you have a medium-format attention to detail, but want the flexibility of a 35mm system.
(Learn more about Kevin Ebi's newest book, Living Wilderness, the first comprehensive portfolio of his fine-art images and download a free preview. Follow his photography on Facebook, Twitter, or Instagram .)
Excellent, thank you for the comprehensive and intelligent review.
Excellence review of the Canon 5DsR Kevin, I too have this camera (got it when it was launched). The massive megapixel is very unforgiving on lenses and camera shake.
I have problem with the Zeiss 35mm f2.0 where at its best performing aperture at f8.0 produces soft images. After reading and understanding of pixel pitch do I know many lenses are manufactured for pixel pitch around 6.0 micron. The 5DsR has a pixel pitch of around 4.3 macron which is very demanding to old lenses.
I am actually looking at the affordable Canon 100-400mm IS L II. Have you gotten any images with the Canon 100-400mm IS L II and how acceptable is it to the Canon 5DsR?
Thank you for your advise,
I'm actually very happy with the performance of the Canon 100-400 Mark II on my 5Ds R. I've used the combination several times and the resulting images are quite sharp. It does get a bit soft with the 1.4 teleconverter attached; however, I still wouldn't hesitate to publish or print even those images.
Post a Comment