We’ve found the R5 to be – by far – the best camera we’ve ever used for stills, and the one that takes the best quality video (requiring me to use my external recorder if done for any serious length of time).
Click on a tab below to get a run-down of the feature’s performance; known issues; hacks; and comparisons with models from other companies.
The Big Picture:
– Long-form video shooters lacking external recorder
Major Capabilities; Issues; Discussions
The R5 sits in the hand a little easier than the EOS R, and certainly the RP, if you know and like the DSLR 5 series. It is, however, just about as heavy as the last 5 series DSLR, the 5D Mark IV, which might be a little more than is optimal (1.6 pounds). The controls are intelligently placed and feel very much like a Canon – in a good way. And it is zippy in almost all ways, which is a big deal when it comes to perceptions of interface quality. It hits you right at the start. The on switch gives you a camera ready to take a picture in 0.4 seconds. The Sony A9 takes 50 percent more time, and – oddly – the A9 Mark II takes 4 times as long. You notice it. The R5 feels like the old DSLRs, which didn’t need any fancy firmware boots to do what you needed to do.
The Theme: Speed
The back screen is maximally used for control as well, even more effectively than the previous R models or the M models, due primarily to a zippy processor that eliminates what occasionally had been glitchy response lag. Moving the focus point with your thumb on the screen, without even looking away from the viewfinder is finally the magical experience it was meant to be. The M cameras almost had it. Sony, eventually, almost had it. Canon finally put all the pieces together.
Gone is the track bar formerly seen on the EOS R that never quite made our lives easier due to – again – a slowness to be responsive. We wonder if, with the new processor internals, the track bar would be a success in that alternative universe. In its place, though, is the most responsive joystick we’ve used on a camera. There it is again, the responsiveness of the controls is a detectable characteristic down to milliseconds, and this R5 has got the power to make this feel really, really good.
The R5 now sports two card slots – a legacy SD card (finally UHS-II) and the new CF Express card format, which should take over the world shortly. Yes, that’s another card format you need to sink half the cost of a camera into, but this one is the forever card. We think. We can be certain that it’s working, at least. The CF Express cards are taking data so quickly off the R5 that it runs 10 frames per second without a buffer. So after you fill the buffer shooting 20 FPS with the electronic shutter or 12 FPS with the mechanical shutter, you still have 10 FPS as the punishment for having been so gluttonous with the shutter button. [See the part a few sentences back about needing to sink half the cost of the camera again into new memory cards.] Because tests run by Camnostic on four brands of cards (more later) showed that most are giving only about 60 percent the sustained write speed the standard allows for, we expect new cards to come to market that will at least outpace the mechanical shutter. Imagine, we are at the cusp of an era where wildlife shooters just don’t care a fig about buffer capacity. People taking up photography today will look at us quizzically when we ask what the buffer capacity of a new camera is five years from now. Boomers, all of us.
Canon always does a good memory hatch – at least with their higher-end models. A great memory hatch can be fully operated with one hand. We can open, remove a card, and shut the hatch in about 1/3rd of a second. It’s a beautiful, human design that is obviously unappreciated by other manufacturers. Sony has begun to improve theirs deliberately over the last round of ergonomic reforms. Most others appear to be two-handers.
The dials are deliberately segregated into three pieces of real estate. You have your front dial running vertically, designed for the right forefinger; the top back dial, oriented parallel to the floor, designed for the right thumb, and then the back dial, oriented parallel to the screen and designed for the thumb as well. This mentally segregates these dials in one’s mind by location and by axis. This is the sort of touch Canon uses to make its ergos just feel right. Panasonic copied a good deal of it over time, culminating in the S series, which is the other camera series with a similarly-satisfying feel. The uneven topography of the top plate shows this design philosophy. The front dial lives on its own plateau, with the shutter button falling forward, not just on its own plane, but on one whose tangent is precisely the angle of a shutter finger pulling itself into its palm. The lizard instinct of a shutter finger is to pull on this axis, and Canon angled the plan of the surface on which the button stands with this in mind.
The receipt of two R5 cameras coming in early August is good timing for autofocus testing, as here in Vermont the nighthawks tend to move up the Connectcut River valley and provide a beautiful, frustrating autofocus and tracking challenge. Nighthawks are aerial insectavores, so they constantly swoop in unpredictable vector changes, typically about 100 to 200 feet in the air, and typically in terrible lighting situations, just at dusk. Perfect.
Shooting them requires very long glass, so autofocus difficulties are compounded by having a bit of a soda straw effect. AF needs to happen quickly, or the bird will be out of the frame before the camera starts to track.
Last summer, we tried out the Sony A9 and A7R Mark IV, the Panasonic S1R and the Canon 1DX Mark II. Of all of those, the A9 was the clear winner for tracking and autofocus. With this sort of a target, tracking is very much linked to autofocus capabilities, as these birds are impossible to hold on a static autofocus point. If you don’t get initial AF, you won’t be able to track, and if you don’t get tracking, you’ll fairly instantly lose AF. But the A9’s resolution is a weak point, and this is the sort of reach-limited wildlife photography that often requires hard cropping. My favorite subjects captured were by the A9, but my favorite final images wound up being from the A7R Mark IV, heavily cropped in.
After three nights of shooting the birds this week, we can definitively proclaim the R5 the top of this camera pile. The autofocus is as fast as the A9 and A9 Mark II, but the tracking is – surprisingly – even better. I never imagined that a camera would be able to interpret through 840mm of focal length a bird 200 feet up in the air twirling and be able to put a little box on the eye of the bird so as to autofocus at just the right plane of focus for that creature in that position at that moment. The deep learning technologies available now in cameras are already surpassing not just our human capabilities, but also our very expectations of what might be possible. This is one of the reasons why the introduction to this R5 crowdsourced portal spoke of this being a generational camera release.
The earlier generation of Canon tracking would focus on people, bodies, heads and eyes, but not do terribly well with inanimate moving objects that didn’t trigger one of those tracking responses. This was a clear advantage of Sony’s, which would allow you to define by focus point what to start tracking, and then stay on it like a dog with a bone. With this release, Canon can provide the same functionality. The “stickiness” of the tracking of inanimate objects is not as strong, though, with default settings. You must goose these up in the settings to get the sort of obsessively-determined tracking as you see on Sony with non-living things.
Relative to previous Canon models, like the 5D Mark IV and the 1D series, and any of the previous R models, the R5 is a clean break, but for the liveview mode of the 1DX Mark III, which is very similar, but an awkward way to shoot wildlife, requiring that one hold the camera out to view the screen rather than look through the viewfinder. Aside from the leap in tracking, the most notable new capability is the ability to keep on a subject without getting suckered into jumping to a complex background. Moreso than with Sony cameras, Canon autofocus would work great for a nighthawk against the sky, but as soon as one dipped below a treeline, you’d expect the autofocus to jump back to the background. Not so anymore. Really this is a function of how much bird is in the frame versus the background.
Estimating roughly, the earlier DSLR autofocus systems would require a bird take up at least 10 percent of a frame in order for they AF to reliably stay on the bird and not slink off to the background. The R5 was holding birds steady in tracked autofocus against contrasty backgrounds while an imaginary rectangle around the birds comprised only 1 percent of the frame or less. The shot at left was taken at 840mm from a half mile away. There are four birds in the picture. The camera was tracking the one at the bottom middle. It is in perfect autofocus in this easier situation with a relatively low contrast background. Counting pixels, this bird is about 0.1 percent of the frame. Of course, there aren’t a lot of pixels on this bird, as it’s too far away, but this is just to show that autofocus is no longer the limiting factor.
If given the choice to use the $6,500 1DX III or the $4,000 R5 for a day of bird shooting, the R5 would be the clear choice, for autofocus and for resolution and cropping ability.
One limiting factor can be the lenses, which have varying autofocus motors. The R5 is going to tell the lens where to put its elements to get the right focus, but the lenses determine the speed at which that is done. Testing with the 800mm f/11 and the 600mm f/11 showed that they were much too slow to handle the tracking of such fast-moving and unpredictable flyers.
Frames per Second
Mechanical, 1rst Curtain & Electronic Shutter
|Battery falling below 70%||1/4|
|Lens aberration correction||1/5|
|Use of grip??||?|
|Wifi use (surprising!)||1/4|
|Auto Lighting Optimizer (ALO)||?|
The above list is almost certain incomplete. To note errors and omissions, please post the the forum linked below.
|Factor||Post-Buffer FPS Hit|
|RAW vs JPG/HEIF||1/2|
|Dual Pixel RAW vs RAW||1/2|
|High ISO (>6400)||1/5|
The R5, like almost all cameras, writes files first to a memory buffer, and then offloads that buffer over time to the memory card. That last process is the slowest part of the process. This structure allows for faster picture taking until the buffer is full, and then the camera will be limited to the write speed to the memory card. This camera is fast enough in its file writing that if you are shooting JPGs, the card writing can be faster than the camera can even collect information, so there are essentially no limits *if* you are using a fast CF Express card. If you are using a slow SD card, there could be any degree of delay, depending on the sustained write speeds.
In-Body Image Stabilization (IBIS) and Image Stabilization (IS)
|Lens||Lens IS||With R5 IBIS|
|RF 24-70 f/2.8||5||8|
|RF 24-105 f/4||5||8|
|RF 24-70 f/4-7.1||5||8|
|RF 70-200 f/2.8||5||7.5|
|RF 15-35 f/2.8||5||7|
|RF 35 f/1.8||5||7|
|RF 24-240 f/4-6.3||5||6.5|
|RF 100-500 f/4.5-7.1||5||6|
|RF 85 f/1.2||–||8|
|RF 28-70 f/2||–||8|
|RF 50 f/1.2||–||7|
|RF 600 f/11||5||5|
|RF 800 f/11||4||4|
High ISO Performance
ISO performance is often mischaracterized as good because the people doing the testing use well-lit situations. In point of fact, photographers use high ISOs in light-stressed situations, and in those situations, the ISO effects on image quality tend to be much worse than in well-lit situations. The best way to test for high ISO performance is to shoot a moving subject at a fast shutter speed, forcing a higher ISO and an underexposed shot. That’s the real world. It’s that adrenaline-filled moment when you know you’re compromising between speed, light and ISO sensitivity.
In that real world, I’d rather have the R5 in my hands than any other camera. The shot below was taken at 12,800 ISO.
This shot of a great blue heron could not have been taken by my Sony A7r IV (AF would not have tracked it adequately); or by my Sony A9 II (its low resolution wouldn’t have allowed me to crop in this far, and the AF may or may not have tracked the bird); or my Panasonic S1R (autofocus); or my Canon 1DX III (resolution). But all those issues aside, I don’t think the ISO performance would have been as good if all images were downscaled to a common resolution.
The R5 is the first camera I’ve ever set the AUTO ISO maximum to 25,600, knowing I could get keepers. Just a month ago, my daily shooter was the Sony A7R IV, and I had the maximum ISO set for 6,400. I believe that the image quality I get at 25,600 on the R5 is the equivalent of the Sony A7R IV at between 6,400 and 12,800; and that were I to process the Sony files to equivalent R5 resolution (the Sony has 30 percent more pixels), the improvement in the R5 would be roughly half a stop.
The Sony files do appear to hold up better to more radical levels adjustments in post processing. Files – to my eye – tend to get a bit crunchy after moving exposure more than three stops on the R5, where the Sony files typically can be moved 4 stops without similar negative effects. The upshot of all of this is that if you require four to five stops of exposure correction, the Sony files may be just as good, but in almost all other cases, the R5 files will get you better image quality.
The R5 has plenty of different ways to control ISO. You can use the multi-controller, set it to any of the dials, use the Q button, etc. The annoying part is that when you use the Set button, the lens button, or the depth-of-field preview button, you can move the ISO up or down, but you can’t set it back to Auto. This has been true of previous bodies, and it’s very annoying, but the R5 does default it back to the original setting, after the next series of shots is fired, which is a big improvement. Where those ISO switching methods used to only be useful to people who didn’t use Auto ISO, now it can be used by the Auto users who intend to make temporary changes to the aperture and then let it fall back to its original setting after the shot.
It should be noted that the RF lenses (and most adapters) have a new control ring on them, which allows them to be set for any number of functions, including an ISO dial. There is a charm to having them set to aperture, as it harkens back to the days when the apertures were manually set on the lenses.
All of the elements in these various tabs greatly affect image quality, but the combination of all of them generate a rough sense of what you’ll hear fuzzily termed “IQ.” Aside from the factors specified in these tabs, there are a number of potential IQ killers:
- Anti-aliasing filters that are too strong
- Dynamic range limitations
- Color rendition
- Lower color bit depth used under processor-stressed situations
The R5 does have an anti-aliasing filter, but it is not strong. By way of comparison, the 5DsR does has a negated AA filter and 5 additional megapixels of resolution, but the R5 is demonstrably sharper than the aging high-megapixel DSLR.
Canon seldom tells people the expected dynamic range of a new camera, but this time reps were telling some reviewers that it should wind up being about 12 stops. Bill Claff, one of the industry’s data saints, keeps records of actual calculated dynamic range data for most cameras, and the R5 will in fact sling just under 12 stops – a fantastic showing besting both the “flagship” 1dX Mark III and Sony’s A7R Mark IV, probably the two cameras most often considered as alternatives. The R5 has slightly more than a full stop of additional dynamic range at base ISO versus the 5D Mark IV and EOS R, which share a sensor. At higher ISOs, there is little or no improvement.
Canon has always enjoyed a reputation for good colors, which probably has more to do with some foibles in competing systems. Tony Northrup did one of his wonderful big data articles, polling his audience to see which pictures they thought looked best, without telling them which images was from what brand of camera. In that blind test, people really didn’t show the preferences that they thought they would. However, having shot Sony for more than a year, we can say that the auto white balance in those cameras is poor in the first few milliseconds of a scene change, and gets better as the frame is allowed to settle before firing off a shot. This sort of speed issue may cause shooters to perceive color problems rather than AWB speed problems. If Tony’s sample images were carefully taken shots, they would not have shown these operational issues, so it would skew preferences. All of this is to say that the color rendition is as nice on the R5 has it has been on previous Canon cameras and that the reasons behind certain preference perceptions haven’t been adequately teased out yet.
Color Bit Depth
The Canon R5 generally shoots in 13 bit color when running at its highest speed. When the camera is taken off the H+ settings (either in mechanical or electronic shutter modes), the bit depth goes up to 14. This is an interesting, perhaps innovative compromise relative to the Sony A9 systems, which received criticism for dropping from 14-bit to 12-bit color when shooting at the fastest speed. The upshot of that controversy was that people generally couldn’t tell the difference, although our own tests showed that – if you tried very hard – you could. The difference wasn’t enough to take the hit on the frames per second. With 13-bit color, it will be interesting to see if it can be noticed at all.
It’s a bit odd that Canon didn’t lead with the lack of rolling shutter – caused by a massively-improved sensor readout architecture – as a key feature. This improvement is the difference between a camera that takes nice still lifes and a camera that can be used for sports, wildlife, weddings and any other genre where there is subject movement. The picture below shows a 2010 Eurocopter 135 P2 helicopter ambulance whose rotor speed at the tip is typically around 450 miles per hour. We were taking shots of a speeding dog chasing around the yard, and a boy on a bike when this aircraft happened to fly overhead. If this rotor doesn’t show a great deal of rolling shutter, your sprinter isn’t going to either.
Rolling shutter – sometimes called the “Jello effect” – is where a slower sensor readout causes a distortion effect that gives an image a fun-house mirror effect, typically smudging the image to the left the further down the sensor you go. Our first impression is that the R5’s resistance to this effect is roughly on par with the industry’s current rolling shutter champion, the Sony A9 II.
There is more to the power features of cameras nowadays than just how many pictures you can get out of a battery. Brute force power is important, but also important is the flexibility of power sources, and increasingly, how software reacts to different situations when the camera is powered externally. But first things first…
The new LP-E6NH batteries have about 1/7th more power than the earlier LP-E6N batteries, now making it competitive with the Sony Z batteries, if just shy of the Z100’s rated power. These new batteries come new with no charge, which may suggest a chemistry change, as Canon previously shipped them with about 60-70 percent power.
The camera is rated to 320 shots per battery, which is of course horse hockey. CIPA ratings are notorious for underestimating actual field performance, as they require much in the way of viewfinder use, post-shot viewing, etc. A rough rule of thumb is to double the CIPA rating. Still, the battery performance is less than that seen with the DSLRs running the LP-E6N batteries by a good margin. Our own performance testing found that the R5 showed inordinate battery usage when we were taking occasional shots, but quite good battery usage when we were shooting at high frame rates. The sample set of two R5 cameras were set to maximum battery draining features, such as the higher viewfinder frame rate. This strongly suggests that the overhead of running the viewfinder and spending more uptime per shot took a toll on the battery. In one instance, we got more than 3,000 shots off of one battery, but that was doing shutter speed tests, involving very large numbers of shots over a short period of time.
A sample set of eight LP-E6NH batteries was run through a Dolgin charger that has an MAH measurement feature. All batteries charged up initially to between 2080 MAH and 2100 MAH. Interestingly, subsequent charges reached lower power levels, but then further chargings brought them back up close to their original spec-busting levels. This is notable in that it is likely that many reviewers will start testing a camera in earnest a few charges in, and that is when they’d get the worst battery performance. It is quite possible the variance we see in opinions on the battery across reviewers may simply be a function of how much their battery happened to have been broken in.
Battery chemistry is a series of compromises, and to all of our tests needs to be added a caveate: the long-term performance of a battery will show itself after a few months of use. The classic mistake many make is to buy third party batteries (nothing against third party batteries) and post an opinion on it after a few charges. They do fantastically. Six months later they will often perform about 80 percent as well as the Canon batteries. So this section will be updated to ensure we can detect any trends.
Speaking of third party batteries, the LP-E6NH does still more communicating with the body, and there are certain functions – some important ones – that work only with Canon’s own batteries. There are not yet any third party versions that claim to immitate the NH status and make the camera provide full functionality with, for instance, maximum frame rates.
Old Battery Performance
A hodgepodge set of LP-E6 and LP-E6N batteries – the older ones included with Canon cameras going back to the Cretaceous Period – were run through the R5 cameras. A few things to note:
- The shots gotten per battery were – not surprisingly – directly correlated to the MAH levels those batteries showed
- The earlier batteries – the oldest of which was 10 years old – performed shockingly well for their age. In fact, some of these batteries were dug out of a drawer that hadn’t been opened in 15 months. One of the older batteries had a charge approaching 70 percent.
Using these older batteries will put the training wheels on the camera for H+ shooting, the fastest frame rate capacity. This will move your FPS down to 15 from 20 when using electronic shutter.
USB power is a big deal for some, but there’s USB power, and then there’s USB power. At its most basic, a camera will allow charging of an internal battery when a USB battery is hooked up to the USB port. The Canon R did this – a first for Canon – but it did not all for the operation of the camera under USB power. This precluded the camera from being used as a “remote” hooked up to a big battery, or to be set up in a studio with a larger battery to prevent the hassle of changing out LP-E6Ns. This could be accomplished by using a dummy battery that is then hooked into a battery, but those arrangements more often than not had their own limitations, especially when using USB batteries. Specifically, those batteries tended to turn themselves off after some arbitrarily firmware-determined time, rendering a remote camera useless unless it had been triggered within a specified period of time. Workaround included a small number of batteries that had firmware that did not turn itself off (expensive and quickly drained) or using a product like TetherTools Case Relay device, which had its own hardware that sipped energy from the battery and employed a roughly 1,000 MAH battery internally to keep a larger USB battery on its toes.
The R5 allows for both charging and running the camera, and this is a bit of a game changer for quickly creating reliable remote setups.
There are two grips designed for the new R5 and R6, the BG-R10, a basic $350 grip with two battery slots, and the other, the WFT-R10A, that costs 2.5 as much and includes faster networking features as well as a rubber interface to allow for a dummy battery to be employed.
Canon caveats its frame rate statistics indicating that they may be affected negatively with the use of a grip, but we cannot seem to make the grip by itself cause a degradation in FPS. The networking version of the grip can of course enable the use of a faster wifi connection, but the wifi tax imposed on FPS seems to be the same with or without the grip. We’re betting there are readers who have discovered the subtleties of this concern and look forward to hearing more on it.
Because the amount of juice left in a battery helps determine FPS, it would be optimal if the two batteries in a grip drained in a pattern where one drained to 70 percent, and then the other started draining. This would maximize the times only one battery needed to be removed for recharge, yet keep the highest framerate for the longest time. A pattern very similar to this has been observed with the WFT-R10A, but we are in the process of verifying with actual FPS data.