Camnostic updated its comprehensive Canon EOS R3 review geared to help R5 and 1DX III shooters make upgrade decisions. Recently added are video aspects and comparisons to the new Sony and Nikon flagships, as well as the R5C. There are a lot of “ifs” and “buts,” but the upshots:
Should 1DX III Shooters Upgrade?
In short, yes – if you think 30 frames per second, more megapixels and significantly improved ergonomics matter enough to cover the <$1,000 difference between a new R3 and a used 1DX III.
Should I Upgrade If I Own the R5?
Probably not, but maybe. The extra $3,000 would buy you a camera with about half the resolution. The R3 does beat the R5 significantly in super-low-light circumstances. Its autofocus capabilities are better, but not demonstrably in most situations.
When the Canon CEO Fujio Mitarai was asked what worried him most in an interview several years ago, he said Canon was too slow to bring products to market, and the market was expecting things to move faster and faster. He was going to light a fire under the product managers, and Canon was going to have a comprehensive constellation of products put out much more quickly that would best all competition. Everything he said mostly came to pass, except the part of it being done quickly. But here we are, at the end of that moonshot, staring at an R3.
We spent all of December shooting this little monster, notching up about 30,000 exposures under sometimes trying conditions, including rain, extreme cold and long swathes of time at the in-laws. We then had the opportunity to compare that experience to varying degrees with the Sony A1, Nikon Z9 and the Canon 5C in the six weeks since. The camera is in many ways the best Canon has ever produced. Aside from limited resolution relative to the competition, it delivers images that other cameras often can’t capture. Below is a picture of the EOS R3 feeding on the carcass of the older 1DX series (taken by an R3).
This article aims to arm existing Canon shooters with the right information to decide if the R3 is a useful upgrade.
We can winnow out a few R5 shooters immediately, for lack of 8K shooting. The R3’s resolution is just too poor to manage 8K video. Of the flagship cameras compared in this review from Sony, Nikon, and Canon’s semi-flagship R3 and R5 (reviewed here) and R5c, only the R3 lacks this capability. Look elsewhere.
For most video needs, though, things are a bit more complicated, so we set out some direct comparisons below in the video section.
Another factor that likely won’t make a difference for most people: build quality. All of these cameras accepted abuse without incident. And we didn’t coddle the camera….
WHY YOU’D BUY IT:
– Reliable 30 frames-per-second (FPS) speed
– Better high-ISO image quality
– Fast, stacked sensor for largely eliminating most shutter roll
– Eye control for moving the autofocus point with your viewfinder pupil
– Low light autofocus down to -7.5 EV (with caveats)
– Flash works with electronic shutter, unlike previous models
– Flippy screen included despite their odd association among pros with amateurishness
– In-body image stabilization
– Best ergonomics of any Canon camera to date, while shedding weight and size
– Better battery
WHY YOU WOULDN’T BUY IT:
– Resolution is poor relative to other mirrorless offerings, including Canon’s own R5
– Related to the above, if you need 8K video shooting, you’re out of luck
– Users married to the screen touch-and-drag method of autofocus point control will find it retired in the R3 (replaced by 2 probably better ones)
– Size is still much larger than an R5 class body, despite being <300 grams heavier
UPSHOT REGARDING PRIMARY SPECS AND PERFORMANCE:
– Low-light performance is excellent. In part due to actual photographic dynamic range sensor improvement, and also likely some software cooking, the images coming out of the R3 looked to be roughly 2/3 to a full stop better than the R5 in image quality at the highest ISO native to the R5 (25k). This is unexpectedly good, especially coming from Canon’s first stacked sensor, which typically would sacrifice dynamic range for faster readout. This comparison was done by taking an image of the same object in light that was two stops below ideal versus the settings and comparing a down-sampled version of the R5 photo to allow that camera’s superior resolution to help compensate. This is likely the most important unsung benefit that the R3 camera can uniquely provide.
The picture below was shot at 50k ISO in the gloaming of a winter’s evening, and then cropped. The R5 can do this if it has about a stop more of light.
– Handling is far better than the 1- or 5-series. Size, weight and button positions are improved and simplified, and it’s enough to make a difference. There are some oddities inherited from the 1 series that were clearly brought over to make the 1DX III emigrees feel more at home, such as the top left duo of buttons, which is essentially the old way of handling what nowadays is done more elegantly and completely with the M-fn button. But there is a busload of 60-year-old photographers who would go apoplectic at the notion one is a replacement for the other.
– The dynamic range is fantastic, better than any Canon camera and on par with the competing Sony A1, slightly better than the new Nikon Z9. This is a great achievement given the introduction of the stacked sensor, which in some ways works against dynamic range. The excellent site PhotonsToPhotos.net is a wonderful source of independent measurement information, and also shows that the R3 noise performance benefits from some algorithmic cooking – something common today among professional bodies. The improvement shown over the R5 and A1 at ISO 1200 and above appears to be part of the “look” Canon bakes into flagship camera software rather than a third native ISO.
– The viewfinder turns on more quickly than other Canon mirrorless cameras.
– Rolling shutter is less of an issue, especially for stills. The stacked sensor is doing its job.
– Where ISO 25,600 was previously the point at which I stopped Auto-ISO from creeping further up on the R5, the R3 takes usable shots at 51,200. There really is almost a stop of better low-light performance.
– 30 frames per second starts to head down the slope of diminishing returns, but it’s still nice to have. We found that we preferred leaving it on the fastest frame rate because the additional frames provided some important shots, despite the nightly horrors introduced in culling.
– The sensor is disappointing in resolution, but images resolve better than the 24mp would suggest. Part of this is an average improvement in autofocus, and part of it is that this R3 contains the same file cooking present in the 1 series, which bakes in some noise reduction and very selective sharpening. Some reviewers indicated the R3 files seemed about as detailed as the 30 megapixel 5D Mark IV files, and that feels about right. Where it falls short with some frequency is disallowing the relatively extreme cropping we could get away with on R5 50 megapixel files. For reach-limited photographers, like many wildlife shooters, this poses a large compromise.
– Auto white balance (AWB) may have improved radically in the past 5 years, but there were still mixed light situations that stumped AWB in the R5, with this reviewer’s multi-hued dining room lighting as one example. No more. The R3 calculates its way out of that orange/yellow/green trap with grace and speed.
– The battery performance is quite good, reminiscent of 1D series expectations. In one instance, we were able to get 16,000 electronic shutter images in one battery charge, but this was taking hundreds at a time for a review of of CFexpress Type B cards. In more difficult, battery-draining circumstances, we would expect between 2,000 and 3,000 shots reliably.
– CFexpress throughput is slightly increased. Most cards show a 2-6 percent increase in data flow over a 30 second burst period. Nothing to write home about.
– Buffer unloads more quickly.
– Allows multiple frames per second settings while in electronic shutter, the lack of which was a frequent complaint for the R5.
– Added a full frame zone for autofocus selection.
– Zone autofocus can now be combined with subject tracking; much like spot focusing can now be combined with tracking.
EYE CONTROL/PUPIL TRACKING:
This is the new sizzle feature for the camera, and it can be useful… for some things. Pupil tracking, called Eye Control by Canon, senses where you are looking via sensors that are facing the photographer. They see where the shooter’s pupils are pointed and will change the autofocus point to that location when a button is pressed. It just takes a couple of quick calibrations and some getting used to.
Some report that it requires frequent calibration, and still others report it doesn’t work well at all with their particular eyes or eyewear, but for most, this is nice to have. We found it useful especially for multi-subject shots, where we wanted to get shots of different individuals on the fly. This would be very useful taking candids at a wedding, for instance, where the focus point would fly from subject to subject. Some sports might lend themselves to it as well, although we generally preferred the smart controller for moving the focus point in a general area and then letting eye detect take over for those situations. The pupil tracking seemed to respect the spot the photographer was looking at for a bit longer before letting eye detect take over for tracking, which was sometimes frustrating during fast action.
Over the Christmas break, we took the camera to a marsh and photographed a group of a dozen tundra swans arrayed on a wide creek at dawn. It worked great for changing compositions with the changing light, and focusing on different birds for each. In particular, it was useful when we wanted to do a focus stack of all of the birds, allowing us to get an image of each bird individually very quickly, perhaps at twice the speed it would have taken with any other autofocus point selection method.
But pupil tracking doesn’t work at all if you’re using the flippy screen, of course, as your pupil won’t be up to the viewfinder. When getting low to the ground, or putting the camera high up, above a crowd, the flippy screen lets you get shots you otherwise couldn’t achieve… but you’re likely going to be using the smart controller to change autofocus points. That isn’t a great sacrifice, but it takes some getting used to swap from one method to the other on the fly.
And that complexity is probably the biggest downside to the eye control/pupil tracking feature. When you have it activated, it can make the viewfinder quite busy. You might have the blue squares showing focus area, as well as a white square indicating the current eye tracking, and then on top of all of that, a yellow graphic might be flitting around, showing where the camera thinks you are looking. As two of these three will be moving in different directions, it can be quite a lot to take in. We wound up using one of the front customizable buttons to turn it on and off, so we could surgically use it when it suited, but otherwise not have the “wandering eye” flying about the viewfinder.
Calibration is simple and quick, and you can get a more sturdy calibration by going through multiple iterations, especially using different light conditions and both vertical and horizontal orientations. That said, when going into a very different environment, we found that we would often need to recalibrate to get accurate placement. The most difficult circumstances in which to achieve accuracy happened in awkward shooting conditions, where the photographer’s face couldn’t be exactly square to the viewfinder, such as when stretching on tippy toes, or quickly rotating the camera to capture something other than a simple 90-degrees-from-level angle.
When re-calibration proved needed, the tracking would often just not quite reach the desired target. To get the shot, we would sometimes deliberately look beyond the actual subject to coax the yellow icon over the desired target. This sometimes caused an interesting feeling, as though we were trying to look indirectly at people, as though we were avoiding their eyes. Re-calibration is pretty quick. We timed it at 10 seconds once you know where to jump into the menu.
This is generally a better video camera than the R5, *if* you don’t miss 8K shooting. The reality is as of this writing that few people are using 8K regularly. Aside from major commercial productions, 8K is used by people wishing to give themselves more zooming and panning flexibility in editing, or for “future-proofing,” which sounds good until you own a closet full of hard drives.
The R3 shoots a beautiful 4K video that downsamples from the whole sensor. That whole sensor isn’t nearly as large as the R5, which will also give you downsampled HQ 4K video, but in looking at side-by-side comparisons, we could not make out any benefit the R5’s additional resolution provided in that format. The big difference is that the R3 doesn’t require an external recorder to not get tripped up by heat-caused interruptions.
This calculation becomes a bit tougher when comparing the R3 to the R5C, Canon’s new video-focused redo of the R5. Heat is no longer an issue with the R5C, which includes active cooling. The R5C also throws in the Canon Cinema EOS interface, with all the tools and tricks enabled for their Cine line, where the R3 is left with the more traditional DSLR-borne hybrid interface. The R5C allows you to boot into either the Cinema EOS interface or the normal EOS interface, depending on the work at hand.
The R5C’s signature feature is the ability to shoot 8K 60p, allowing for both zoom-and-pan editing in post while exploiting slow motion.
A touted benefit of the R3 has been the stacked sensor improving the rolling shutter effect. This is true. But the R5’s rolling shutter was relatively well controlled, especially for a full frame sensor. Unfortunately, the R5C didn’t provide any benefit over this.
Perhaps the biggest R5 handicap is the arbitrary recording limit when making videos while recording internally. That 30 minute limit is lifted for both the R3 and the R5C. For high bitrate video formats, that arbitrary limit wasn’t as much of an annoyance as might be expected, given that the heat limitations of the R5 meant that shooting 8K or 4K HQ for 30 minutes was already an unlikely prospect. But people conducting hours-long interviews or filming documentaries found this maddening, as it affected all video formats in the R5.
The R5C and the R5 suffer from using a micro-HDMI port – fine for occasional use, but a liability for frequent or rough use. None of the other cameras considered in this review limit output to that weak, lilliputian format.
The upshot for video is that the R3 is far-and-away better than the 1DX III, and in almost all ways better than the R5 (not for 8K shooting, obviously). But the R5C complicates this choice, if video is the primary intended use. The R5C does not have, practically, a heat limit. It is the only one of these cameras with active cooling. It shoots the sharpest 4K video of the lot, but it doesn’t have rolling shutter as well controlled as the R3, and it suffers from the use of the micro-HDMI port. Because the R5C is larger, heavier, and has ditched in-body image stabilization, it isn’t as good a still choice, so hybrid shooters will likely wish to choose from one of the other R cameras. But if you want 8K 60p video, there is no other choice on the market today.
SHOULD I UPGRADE IF I OWN THE 1DX MARK III?
In short, yes – if you think 30 frames per second, more megapixels and significantly improved ergonomics matter enough to cover the <$1,000 difference between a new R3 and a used 1DX III.
If you shoot low light frequently, and you need cross points in your autofocus array, the DSLR could prove better for focusing. Mirrorless cameras are good at using phase detection in one axis to determine autofocus, but not in both axes – for lack of cross points that DSLRs use. The R3 will sometimes recognize an object as having been selected, but then it will ooze over to the nearest vertical line, which might be close piece of the background. Then that selected column might ooze over to the side of a building because it has more contrast. This doesn’t happen in most cases, but in rare low-light circumstances, it can be crazy-making.
SHOULD I UPGRADE IF I OWN THE R5?
In short, probably not, but maybe. The extra $3,000 would buy you a camera with about half the resolution. The R3 does beat the R5 significantly in super-low-light circumstances. Its autofocus capabilities are better, but not demonstrably in most situations.
We tested this side-by-side, though, with a screaming-fast brittany spaniel running at full tilt toward us. The R5 had a 70 percent hit rate of having the shots perfectly in focus, with those that were out of focus happening primarily when the dog was within 10 feet, and the autofocus had to make the most extreme adjustments to compensate. The R3 missed only one or two out of 50 shots, and keep autofocusing even as the dog passed.
People who need that sort of hit rate known who they are, and might have an R3 over the R5. But, in most autofocus situations, such as birds in flight that were more than 40 yards away, the cameras performed more equally. This was disappointing, as we were hoping that the R3 would show a big improvement, especially in keeping focus on distant targets set against high-contrast backgrounds.
The throw-away images below of an eagle flying away at quite a distance shows a typical moment in which a vertically-oriented background grabs the autofocus.
This is as apt to happen with the R3 as it is with the R5, unfortunately. We tested this with multiple autofocus settings, which didn’t appear to help significantly.
The R3 does have 50 percent more frames per second, which can be a big deal for some users. Buyers are definitely going to want to find a good and large CFexpress Type B card for this camera. 30 FPS is addicting. It has a better battery, and better ergonomics and some innovative additional features (particularly eye control and the smart controller). If you find that the R5’s resolution is overkill, and you don’t crop images much, you may indeed find that the R3 works better for you. If you like to take distant wildlife pictures, and you have only a 400mm lens, you’d likely prefer the R5.
Essentially, if money were no object, the decision would come down to whether or not you were more reach-limited or more light-limited in the type of photography you shoot. If light is relatively plentiful and long reach is important, the R5’s resolution may be the best benefit. If, instead, you shoot close-in events in low light, the R3 is likely a better choice.
PROS AND THE TRICKY AUTOFOCUS SELECTION PUZZLE
Whether they like it or not, professional photographers have to play a little video game called “get the shot.” In it, they’re presented with a tiny viewscreen showing a potential image for which they can get paid if they take a picture at the right moment; while balancing four settings to maximize the available light; and moving fingers with lightning speed to make sure the most appropriate focus point is the one that will control the camera’s autofocus function. It’s a lot to think about all at once, especially as most would rather be more worried about anticipating the action and adjusting the composition to make for a better image. Canon has been working hard in the past five years to fix the need for the lightning fingers. First in its 1DX Mark III, with the Smart Controller (a super-sensitive nipple that can very quickly and finely adjust the focus point) and now with the R3, which sees where you are looking, and automatically gives the photographer the option have having the camera focus there.
It’s done a pretty good job with both. The biggest problem now for the photographer is deciding which of several methods to use in the moment for moving the focus point. This too-many-options issue was so great that Canon actually disabled another of its excellent focus point selection methods: using the back screen as a controlling touch screen while looking in the viewfinder. People married to that method in the R5 or R6 may be disappointed to see its absence. But it makes sense. The R3’s screen is further away from the other controls, making this method inferior to the two added ones: the smart controller and eye control.
Different types of photographers operate in communities that have different cultures. Sports photographers and other action-oriented pros pride themselves in the speed with which they can control their hardware to capture subjects often at various distances to them. These subjects tend to conform to the playing field and repeat similar activities over and over again, reducing the finger challenge somewhat. Wedding photographers try to make the equipment disappear and swim among crowds of celebrants in hopes of capturing the mood of the crowd and tenor of conversations. The lightning fingers video game satisfies the sports photographers pretty well, but has always been considered more of a chore for portraitists, wedding photographers and others. Those people are going to like the eye control selection feature, which works great so long as the light environment doesn’t change so much during a shoot that it requires recalibration too often.
THE INTERFACE TEST
There comes a time when a photographer needs to get someone nearby to take a shot with their camera. It can send shivers down the spines of pros, especially those with highly customized gripped bodies with re-mapped buttons. My wife – who has since become a decent photographer – was so used to my little list of instructions, that she’d wave me off saying, “Yeah, yeah, I know, back-button focus, and move the focus point. Sure, sure.” But doing this with a newbie provides one of the best interface tests. This was brought home to me last month when my family visited a tree lighting ceremony with a couple friends. My buddy Tom, a novice, offered to use the R3 take a picture of me and my family together. I’d had the R3 for a day and had two versions of back button focus set up – one with the smart controller, and the other with pupil-directed autofocus point selection. As I handed the camera to him, I contemplated what could I possibly tell him to do to let him take this simplest-of-all-shots? Instead of introducing the concept of pupil sensors, or even what a smart controller was – or, heck, even the idea of selecting an autofocus point – I had to tell him to just fill the frame with us and press one of the buttons first, and then take the picture. He did. The autofocus was spot-on because all of us were a wall of people at the same distance, and wherever the heck the autofocus point was sitting, it would work. Of course, then he took 145 pictures by pressing the button two or three times, waiting for a “click” that would never come because the camera was set to e-shutter.
FOIBLES OF THE R3
What follows is a list of mostly minor annoyances experienced with the R3, some of which may be targets for future firmware fixes.
– The new hotshoe cover is designed to fall off and get lost on day one. Unfortunately, with the new hotshoe capabilities and interface, this is now more of a weather sealing liability than previous hotshoes. At this time, Canon does not sell extra hotshoe covers separately. Our solution to this was a small piece of grip tape fixing the hotshoe cover on. If you own an R3 and are reading this to affirm your purchase decision, stop now and go tape your hotshoe cover. Tomorrow might be too late.
– We felt that when we changed the autofocus cases to be more “sticky,” the results didn’t provide much in the way of additional stickiness.
– The smart controller doesn’t immediately respond to your finger on it. You must first use the joystick component of the smart controller or half-press the shutter button before the smart controller sensor starts working.
– There were times when the autofocus indicated it acquired focus in low light when it clearly did not. The specs list a -7.5 ev as the new autofocusing limit, but we didn’t perceive it to be much better or worse than the R5, which is rated to -6 ev. Our perception was that it was willing to make a call as to whether or not the subject focus was acquired down to -7.5 ev, but whether it was or was not was little different from the R5’s system.
– The ability to use the touch screen to make an autofocus point selection has been removed. We didn’t miss it because of the addition of the smart controller, but others might.
– When reviewing images on the back screen, the smart controller is frequently activated by mistake, making it more difficult to view. It would be better to have to press the smart controller down prior to it operating as an image viewing panning control. Essentially, Canon should switch these behaviors and have it not automatically activated in image view mode, and have it activated initially when used as an autofocus point selection tool.
– The hotshoe cover is a mess. It doesn’t stick on as well as is needed. There appears to be an air gap between the cover and the actual body, but this is just an external flange. We conducted a water test on the shoe cover (see image below), and it indeed passed – so long as the cover remains unlost.
MYTHS TO UNBELIEVE
– The R3 doesn’t improve the performance of the RF 400mm f/2.8 or RF 600mm f/4 in any fashion, other than perhaps driving focus motor a little faster for a longer period with its superior battery, much like the 1DX series does with the EF equivalents. Those RF versions of the big whites do, indeed, appear to be true copies of the older EF Mark III versions with an in-built adapter.
– The Eye Control autofocus point selection feature isn’t the same gimmick that appeared decades ago in the EOS 3 DSLR. This digital version works, and can work well when calibrated. It is true that it won’t be your favorite mechanism to do this across all types of photography, but for some types it was unbeatable.
– The 1DX Mark III lacks a significant superior feature. There are some minor wins for the DSLR. It has a lower viewfinder latency because it doesn’t actually have a digital viewfinder. The 1DX III does genuinely autofocus better with horizontal objects in low contrast situations, but this problem presents rarely.
– This last one is typical for the Canon faithful. They may believe that, because Canon released the R3 with a sensor resolution similar to that of the low-res 1 series, the rumored coming R1 camera will assuredly be a high-resolution monster. Canon might produce a high resolution camera, as it has done before, but to marry one to a gripped body and call it a flagship would be akin to Canon opting to move its headquarters to Miami. It would be a pretty unlikely cultural shift. Because this is what most people want doesn’t mean it’ll be what Canon does.
– In four weeks of shooting with the R3 over perhaps 30,000 images, we experienced two lock-ups, which required battery removal to resolve. This seemed similar to a similar phenomenon some R5 shooters experience occasionally. Some forum dwellers report similar occasional issues. It was not frequent enough to flag as a significant problem, as it is easily resolvable quickly.
– In mist or rain, the viewfinder often thinks droplets on the face sensor are actually a face, and it will turn off the back display and turn on the viewfinder. It seems to have been made more sensitive than the R5, or perhaps the reconfigured eye piece for eye control allows the moisture to collect more quickly.
CONCLUSION: CANON HAS A NEW FLAGSHIP CAMERA
Canon finally made a pro camera that ditched all the irrational elements kept for years primarily so that the male pro photographer could have a camera that emphasized his own manliness. To the degree it was heavy and loud and large suggested the camera needed a “real pro” to handle it. Not only did Canon make a better camera with the R3, which feels better in any gender’s hand, but they added in features that this pro market was reluctant to tolerate; like a flippy screen. Canon finally dropped the macho bull to maximize usability. And you can feel it.
Still today, Canon has been careful to make sure its various national marketing organizations don’t call the R3 its new “flagship” camera. Canon UK even had to take down some language on its website soon after the camera’s announcement to placate the existing 1D shooters whose self-perceived masculinity might be diminished. But this reviewer can tell you – as are all the other reviewers – that the R3 is Canon’s real flagship camera. The 1DX Mark III – Canon’s last DSLR flagship – is inferior in almost all significant ways.
The contenders against the R3 remains the R5 – which is superior for those who value resolution above the R3’s 50 percent more frames per second – and the R5C, which is unique for its active cooling and capability of producing 8K 60p video. For everyone else, the admiral’s colors have ascended the R3’s mast. That resolution difference will also come into play for people considering from among different camera systems. Nikon’s and Sony’s new flagship bodies are high resolution beasts. Nikonians will have a good excuse to stick to their brand, and many rational people are going to choose Sony as well. The R3 is a respectable flagship, but it isn’t a market show-stopper. That said, it well deserves the attention it is getting and is going to sell the bejesus out of whatever’s left of Canon’s supply chain.