So Canon has finally announced their full-frame mirrorless camera, the EOS R.
Personally, I think Canon is dammed if they do and damned if they don’t here. No matter what Canon releases people are going to be disappointed, upset, and more than likely the media is going to come down on them as if they’re the most incompetent dunderheads in the world. This is already the case if you keep up with the rumor mill.
As for me, I’ve said this the past two
rudderless mirrorless rambles that I’ve become less of a fan of mirrorless cameras than I was in the past. A year ago, this wasn’t the case, but in getting back to my roots with wildlife photography, I’ve very found that the necessary power consumption of them is a big negative for me.
Moreover, while I understand the benefits for using the sensor for everything, I don’t see it being nearly as much of an amazing point of awesome that I want to toss my
antiquated junk DSLRs in the garbage and buy amazing new slightly different mirrorless cameras to replace them. Consequently, while I’m pretty sure I’m going to pick up a mirrorless camera sooner or later, I don’t see it as a replacement but as a complement for my existing DSLRs.
Short at the Register
Back in my first ramble, I said I expected what’s now called the EOS R to have a shorter register than the EF cameras, and that’s the case.
There were rumors floating around at the time that Canon’s mirrorless might use the EF mount and maybe some short back focus shenanigans.
As someone who’s in the not giving up my DSLRs any time soon camp, while an EF mount mirrorless would have been in some ways idea for me, the engineer in me never saw that as a realistic path for Canon.
As I noted before, the engineering advantages of the shorter register really are overwhelmingly in favor of the change.
What’s interesting about Canon’s RF mount is that they’re going with a register of 20 mm, not something shorter. Compare this to both Canon’s EF-M and Sony’s E mount, which uses a register of 18 mm, or Nikon’s Z mount that uses an even thinner 16 mm register.
State of the Mounts
With full frame mirrorless options from all 3 major players, now is as good a time as any to look at the state of the mirrorless mounts.
Sony’s E mount is arguably in the worst shape of all three. Originally designed for APS-C sensors, squeezing a full frame chip in there is a bit of a hack. The throat diameter is only 46.1 mm; this is about the same as Nikon’s F mount and Canon’s EF-M mount and only 3 mm larger than the diagonal of a 24mm x 36mm frame. This limits, or at least complicates, the ability to use very fast (e.g. f/1 or f/0.95 lenses) on the mount. While I don’t think Sony will be replacing the E/FE mount anytime soon, it will limit likely complicate their ability to roll out exotic fast lenses in the future.
Then there’s Nikon’s Z mount. With a 55mm throat diameter and a 16mm register, it’s big and shallow, which makes it potentially a prime platform for really fast lenses. In fact, this is kind of what they seem to be touting with their 58mm f/0.95 Noct prime. However, the 16 mm register comes — and Nikon admits to this, somewhat proudly — with the need to design thinner shutter mechanisms, and generally makes it a tighter squeeze to fit anything between the sensor and the lens mount.
That said, I’m not convinced that super fast lenses are a real long term selling point. There’s a lot of flash, and bragging rights, around a f/0.95 lens, but the practicality and utility of something that fast is much more limited. Compounding matters, the current state of optical technology (i.e., we can’t break physics yet), is that these lenses will be very heavy and very expensive.
And this is ultimately why I don’t think Sony’s E mount is that substantially disadvantaged compared to Nikon’s Z and Canon’s RF mounts. Sony can still mount f/1.4 and probably even slightly faster lenses on the E mount, and that’s more than adequate in practice.
Finally we get to Canon’s new RF mount. With a 54 mm throat and a 20 mm resister it’s really somewhere in the middle between E and Z.
The 54 mm throat diameter is only 1mm smaller than the Z mount, and the same diameter as the EF mount. Canon already has done an f/1 lens for the EF mount, back in the late 90s even, so it should be even more practical on the RF mount.
However, over the years, Canon has slowly moved away from insanely fast glass to more reasonable speeds. Their 50mm f/1.0 was replaced by a 50mm f/1.2, their 200mm f/1.8 was replaced by a 200mm f/2, and it very much seems as though their new 85mm f/1.4L will supplant their 85mm f/1.2L in practice too.
Honestly, this kind of thing makes a lot of practical sense. Nikon’s Z 58mm f/0.95 S Noct, is rumored to come with a $6000 street price when it’s released. Likewise, Leica’s 50mm f/0.95 ASPH is a $11,300 lens. Even Canon’s EF 50mm f/1L was $3000+ dollar lens in 1988 (equivalent to ~ $6800 in 2018).
However, the interesting point to me is the 20 mm register.
One train of thought here is that Nikon’s 16mm register might benefit them is in alleviating a need for a retrofocus design with a 24 mm lens, either for a 24-something zoom or a 24mm prime. If you do some quick and dirty math using Leica’s 24mm f/1.4 Summiliux, the rear most parts of the rear element sit about 16 mm in front of the film/sensor plane.
Canon’s 20mm would seem at least on first glance, to preclude that. Of course without a mirror, there’s no reason Canon can’t inset the rear element into the body like their short back focus EF-S lenses do either.
Of course there’s also the possibility that it really doesn’t matter. Either there’s simply not enough room to completely eliminate retrofocus designs for wide angle lenses. Or potentially it’s just not as beneficial to do so. Remember, retrofocus designs have the light striking the sensor at closer to perpendicular and help with vignetting.
What seems reasonably clear, is that a 20 mm register shouldn’t be a disadvantage in practice.
The important points here is to insure that the mount doesn’t appear to be unnecessarily constrained. The mount is what’s going to carry the platform forward. While Sony was able to squeeze a full frame sensor in their E mount, I’m sure if they were to do it over again, they would have upsized the mount a little. Likewise, Canon certainly could have done the same with the EF-M mount, but that too would have been a similarly short sighted move.
In this respect, both Canon and Nikon have played a much better thought out long game than Sony did. Both RF and Z mounts will be something that can carry their respective companies and platforms into the future.
Reticulating Splines Articulating Screens
Articulating displays is something that I’m increasingly convinced that no camera should be without at this point. Even for this jaded photographer who values ruggedness, reliability, and a mirror and OVF, over flashy features, they’re just way to useful in so many situations to ignore.
However, not all articulated displays are created equally.
Both Nikon and Sony have elected to go with a tilt up-down type display on their respective mirrorless platforms. Canon has gone the other way with a side folding design.
Which way is better is probably arguable. Sony and Nikon’s solution yields a is somewhat more compact footprint, though it’s entirely focused on the operator being behind the camera.
Canon’s solution that flips out to the side then rotates 360° from there does increase the camera’s foot print when being used. Plus you can’t just tilt the screen up or down without flipping it out. However, the flip side is that Canon’s screen can be rotated to point forward.
Now you certainly can argue that this is merely a “selfie” feature, and therefor unimportant for the pro/serious user. However, this clearly isn’t a pro mirrorless camera.
Moreover, more and more high end traditionally still cameras are being used for video work. In these cases, having a display that lets you see what you’re doing while in front of the camera is a huge benefit.
Even as a decidedly non-video shooter who has virtually no interest in being in front of the camera, there have been many times where I’ve ended up hooking up my SmallHD monitor to my 5D mark IVs so I could be in front or to the side of my camera and be able to see what it was shooting.
Part of this goes back to what I talked about in the previous ramble about Nikon’s cameras seeming a bit confused. Vlogging is a big thing, and being able to see how your shot is composed when you’re in front of the camera is a pretty important part of doing that successfully. As things stand this year, that’s not something you can do with either an Alpha or a Z camera, but it is something you can do with the EOS R.
Lenses are a big deal. Probably a far bigger deal than the cameras themselves. Starting out with a new platform getting the right mix of lenses out there out of the gate is at least moderately, if not very, important.
Fortunately, with all these cameras being mirrorless designs, this isn’t a clean slate kind of process. In the case of the EOS R, there will be EF to RF mount adapters so the gaps can be covered with existing EF lenses.
That said, native lenses are important to have, and building out the native system needs to be a major focal point for the manufacturers.
In many ways, this was something I think that hampered the adoption of EOS M cameras. Canon simply didn’t seem interested in building out the native lens selection themselves. As a result, nobody else really saw a point in doing Canon’s work for them.
In my ramble on the Nikon Z system, I criticized Nikon for prioritizing the 58mm f/0.95 Noct — a lens I consider only really being there only for bragging rights — over rolling out more broadly used lenses faster. My point being that while sure you could use F mount lenses on an adapter, it’s often a better experience to use native lenses on the platform proper.
In fact, I might even take this argument one step further. Of the 12 lenses Nikon has mapped out going in to 2020, 7 are primes and only 5 are zooms.
Rebagging Cats, Primes versus Zooms
There’s an age old debate in photography over primes versus zooms. One that’s at least worth talking about again in the context of mirrorless cameras.
On one side, the prime users will point out that primes can provide substantially better image quality because they can be optimized for the one focal length that they are. Moreover, the lenses are individually smaller and lighter.
On the other side, the zoom people will point out that zooms offer substantially more flexibility than primes and that this outweighs the differences in image quality.
I’ve fall squarely on the zoom side of the fence, the flexibility is key. This isn’t to say I completely eschew primes, but they’re definitely not my daily driver lenses.
The primes versus zooms situation got real interesting in the early 2010s. At that point, computer based lens design really started turning out some amazingly high quality zooms. Lens’s like Canon’s EF 24-70mm f/2.8L II USM and Nikon’s 14-24mm f/2.8G were redefining image quality at the focal lengths they covered.
However, this also came at a time where a 36 MP D800 was considered an obscenely high resolution camera. As 36 MP has gone from super high end to merely middle of the road, the demands placed on the lenses have gone up as well.
With 47 and 50 MP cameras being more common now than not, zooms are in more and more cases just not providing the image quality they seemed to be good for just a few years ago. As one of my friends found out, somewhat to their dismay, their fantastically tack sharp lenses on their D4s just were cutting it on their new D500s and D850s.
That said, in a lot of respects, I think the cat has already out of the bag. This isn’t the 70s and 80s, zooms with good enough performance are definitely doable and doable with reasonable (for camera gear) prices.
With this in mind, I do question having a heavy focus on primes in spinning up a mirrorless platform. While I wouldn’t eschew primes at all if I was Canon or Nikon, I would definitely focus on covering the highest popularity zoom ranges (e.g., the 24-70 and 70-200) ranges first, before I started trying to roll out showy and niche primes.
In this respect, I think Canon has done a better job than Nikon, at least for the first 4 announced lenses.
Two of these are primes: an RF 35 mm f/1.8 IS STM, and an RF 50 mm f/1.2L USM. The other two lenses are zooms, an RF 24-105 mm f/4L IS USM, and an RF 28-70 mm f/2 USM.
A lot of why I think this is a better opening salvo revolves around the two zooms. The 24-105 mm f/4 is more flexible due to the longer long end than the 24-70 mm f/4 that Nikon rolled out. Put simply, having a bit more range is useful, and it’s part of what offsets the slower f/4 aperture compared to a 24-70mm f/2.8.
On the other end there’s the 28-70mm f/2L. Now this is going to be a big, heavy, and probably expensive lens. But if you’re, for example a wedding photographer, this is going to be a must have.
Finally, there’s the RF 50mm f/1.2L USM. Nikon is basically releasing 2 normal primes in the first year of the Z cameras, a 50mm f/1.8 and the 58mm f/0.95. Canon is releasing one a 50mm f/1.2. The f/1.2 aperture is still extremely fast by any reasonable standard, but the lens comes with a couple of points that I think make it more reasonable. First at f/1.2 it’ll be smaller than the f/0.95 Noct, and lighter. Secondly, I suspect that by not going to f/0.95, owners wont be carrying around extra glass, and weight, that they aren’t fully utilizing. Finally, it’s an autofocus lens, where the Nikon Noct isn’t.
There’s an even bigger kicker here though with Canon’s RF mount lenses, a control ring. Every announced RF lens, and two of the three EF to RF adapters add another control ring to the lens.
According to Canon’s currently available info, this control ring can be configured to control at least:
- Shutter Speed
- Exposure Compensation
I’ve always been interested in the way that camera UIs have evolved over time and adapted to changing technology. Adding a control ring to the lens in some ways plays back to the classic lens with manual aperture ring, but with a whole lot more flexibility.
This is also a point that’s a big differentiating factor between Canon’s system and the competition. More importantly, the EOS R, at least, isn’t eschewing a dual control dial approach and relying on this on lens dial; the new dial is an addition to the system not a replacement.
I know this is going to be controversial, but in some respects the EOS R seems to me to be a slightly less confusingly thought out than Nikon’s initial Z models are.
Of course, I can already hear the cries of Canon bias on this one. When push comes to shove though, the EOS R looks squarely like a 6D like prosumer/serious enthusiast camera, not something that’s attempting be a super high end pro body.
Don’t get me wrong, both the EOS R and Nikon Z are going to find places in pro kits, this isn’t about the camera being more or less capable. What I’m talking about is a consistency in features and capabilities that doesn’t end up being confusingly contradictory.
At a $2300 price point, the EOS R is decidedly in the high end prosumer camera in price range. In some respects, I feel this gives Canon some wiggle room with the specs that I would be much more frustrated with if we were talking about a pro camera.
To start with Canon is using UHS-II SD media in the EOS R.
To be honest, I think this is a much better move in general than Nikon’s XQD. While Nikon seems to be all in on XQD, it’s not something that the industry as a whole has adopted. One company pushing it, is always going to have it on the wrong end of the economies of scale compared to formats like SD, SD Express, CFast, and CF Express.
What I’m sure Canon will get flack for is having only as single card slot. Though again, Canon seems to be positioning this as a higher end consumer camera not a high end pro camera. Moreover, Canon has never put dual slots in their consumer cameras, so the move is at least internally consistent for them.
While two slots would make it more attractive to pro users, at least Canon isn’t going backwards like Nikon did in going from 2 slots to 1.
Again, like usual the discussion at Canon rumors is going bonkers about technobabble when it comes to the sensor. “Oh no, it doesn’t say that it’s BSI.” and so forth. The sensor tech is a means, not the end.
The sensor itself seems to be closely related to, but improved upon, the sensor in the 5D mark IV and the 6D mark II. Hopefully it’s more closely related to the 5D mark IV than the 6D mark II, as the former does get better dynamic range at low ISOs.
Resolution weighs in at 30.3 MP. The ISO range is 100-40,000 with Lo (50), H1 (51,200), and H2 (102,400). ISO 40,000 at 30 MP is entirely reasonably and consistent with either improvements to the 5D4’s sensor’s ISO (base range 100-32,000), or the pixel pitch of the 6D mark II’s sensor. What’s arguably missing, at least compared to the competition, is in body image stabilization (IBIS).
Related to the sensor in part, is frame rates. In a move that I would describe as being par for Canon’s course, the frame rate doesn’t come with quite as much of the *,**, †, ‡, dance as some of the competition.
Continuous frame rates max out at 8 FPS with AF and AE locked. With servo AF, the best case is 5 FPS.
This is definitely not the 12 FPS max of Nikon’s Z or the 10 FPS max of the Sony A7 III.
Is this disappointing? To be honest, I’m not sure.
I certainly would be far more incensed if the camera was aimed at the 5D/pro level. But while there’s always a lot of crowing about wanting more FPS in the various online communities, I’m not convinced of the general need. I owned a EOS-1D mark III when they were top of the line cameras, and with that had a 10 FPS camera when that was fast. What I found is that more often than not the high frame rate just generated more work for me, as opposed to making it possible for me to capture things I previously couldn’t.
I think the more interesting question is whether or not the frame rate cap is a purely technical limitation or a market segmentation one, or a bit of both.
Arguably the biggest redreaming factor for the EOS R is Canon’s Dual Pixel sensor tech, and the AF performance and quality that comes with it.
The camera’s AF range is -6 EV to 18 EV with the 50mm f/1.2 lens. This sounds impressive, but it generally is in line with other AF solutions.
Though honestly, if you expected something truly amazing and not a bit of marketing fluff, I think you’re barking up the wrong tree.
For example, my 5D mark IV is rated at -3 Ev with an f/2.8 lens. With an F/2 lens, it would be good at -4 Ev, with an f/1.4 lens it would be -5, with an f/1.2 lens it would be -5.5ish Ev.
If we compare this to the Nikon Z cameras, they’re rated -1 to +19 EV with an f/2 lens and -4 to +19 EV in low-light AF mode (again with an f/2 lens). At f/1.4 that would be -5 EV, at f/1.2 that would be -5.5 Ev.
Baring a small amount of fudge factor, and assuming that the EOS R is really good for -6 EV at f/1.2, then the AF is rated for a half stop darker environment than it’s contemporaries.
Of course, as I said, this should be expected. There’ precious little reason to push for much lower light performance than that. EV -6 is basically a scene illuminated by starlight only. Moreover, to get better low light performance, you need to either have much bigger pixels or spend much longer exposing them to make your meter reading. In either of these situations, you start making compromises to the accuracy or speed of the system as a whole.
What’s far more interesting than the EV range, at least to me, is that Canon seems to be claiming the the DP-AF system in the EOS R is capable of operating with lens and teleconverter combinations with apertures as small as f/11.
Normally, this wouldn’t be a point worth mentioning. That Canon does mention it is interesting.
Contrast base AF systems don’t have aperture limits at all, they aren’t trying to determine phase from light and they can measure contrast with any aperture size. In a hybrid system, that is one that falls back to contrast AF, there is no reason to talk about aperture limits.
On the other had, it’s a necessary part of describing the capabilities of a phase based AF system.
Fundamentally, Dual Pixel AF is a phase based system, and as a result aperture limits would apply. However, it’s also functionally a hybrid system given that it uses the image sensor and there’s nothing to stop the camera from being able to fall back to contrast based methods when the AF fails.
This could ultimately mean a number of things:
Admittedly, it could just be marketing pointing out something that’s not really material (kind of like how both Canon and Nikon seem to think the number of contacts in their lens mount is a selling point).
It could meant that the PDAF capabilities of DPAF are capable of generating proper phase information down to f/11 without the camera having to fall back to contrast based focusing.
Finally it could mean that Canon is 100% relying on phase based DPAF focusing on the EOS R and that there is no contrast based fallback at all.
We’ll probably just have to wait for someone to stack some TCs and a slow lens to get an f/16 combination and see what happens.
I keep drawing lines between consumer and professional cameras, and this is one of the places where Canon’s design people draw one too.
In cameras like the 1DX and 5D mark IV, their video capabilities have definitely trickled downs from the cinema cameras. For example, 4K mode is DCI-4K at 4096×2160 not UHD 4K at 3840×2160.
On the other hand, consumer cameras, like the EOS M50 that do 4K shoot at the UHD resolution not the DCI one.
This also applies to the EOS R. Its 4K is UDH, not DCI, though it is h.264 compressed not MJPEG. Frame rates go up to 30p at 4K, 60p for Full HD (1920×1080), and 120p for HD (1280×720).
That said, one thing I’ll be very interested in seeing is the comparative quality between 4Kp30 on this camera versus the 5D mark IV. Both will have 500 Mbps recording modes (the 5D4’s MJPEG is 500 Mbps, and the 4K ALL-I mode on the R is 480Mbps). While H.264 is a more efficient codec, my testing has shown that ALL-I H.264 and MPJEG both at 500 Mbps bit rates don’t different by that much.
That said, I can hear the gnashing of teeth over the 500 Mbps files being just as huge as the files from the 5D mark IV. Along with the cries of how could Canon be so stupid. Though practically speaking they aren’t since the IPB 4K mode uses a much more sedate 120 Mbps.
Power and Battery Life
I criticized the Nikon Z cameras for poor CIPA tested battery life, and to an extent the same complaint applies here too.
The EOS R uses the same LP-E6 series battery packs that the 5D mark IV and 6D mark II use. Similar to Nikon’s EN-EL18, these are 14 wH batteries operating at ~7.2 V.
And right out of the gate, the Canon’s numbers aren’t very impressive either. Though they are marginally better than Nikon’s.
However, Canon has clearly thought about power a little be more than Nikon did. Leaked performance data indicates that there are two “modes” for the camera to operate in: “smooth” and “power saving”. Though it’s not clear at this time what the exact difference are.
Again according to leaked performance data, at room temperatures the R gets 370 and 350 frames when the screen and EVF respectively in smooth mode. This is 20-40 frames better than Nikon’s published 330 and 310 frames for the Z7 and Z6 respectively. However, it’s a far cry from the 650 and 550 frames for Sony’s A7R III in the LCD and EVF respectively.
However, setting the camera to power saving mode, ups the frame counts to 450 and 430 shots for the LCD and EVF respectively. This is a lot closer to the A7R III. Though a lot of this is going to depend on just how power savings mode affects the fluidity of the display.
Finally, there’s an eco mode for the LCD only, this ups the frame count to 540 and 560 frames for smooth and power saving modes respectively.
So in the best case, you’re looking at 560 frames out of a 14 wH battery, compared to 650 frames out of a 16 wH battery for the A7R III.
Fortunately, Canon also doesn’t apparel to have totally derped up the release on the power front either. The camera is coming to market with a battery grip, the BG-E22. This pushes the performance up to 700–900 frames on a pair of LP-E6N batteries.
While this still isn’t great in the bigger picture, it’s a far cry better than having to pop out the battery every 300 frames and swap it for a charged one. 700-900 frame is getting on to a reasonable days worth of shooting. Of course, a 5D mark IV can do that on a single LP-E6N, so we’re not talking about setting records here.
When push comes to shove, there’s really no way to get around physics. Mirrorless cameras are power hogs, and compared to DLSRs they always will be. There’s really no way around that. Providing a view to the user is tantamount to shooting video. Even simple things like metering and focusing require much more extensive processing than the simpler dedicated sensors in DSLRs.
I don’t know, I’m torn on whether or not I like the EOS R.
I want to be far more critical of it than I feel I am. I want to say, much like I did with the Nikon Zs, that even though I strongly dislike Sony I still feel that they have a better platform.
At the same time, the EOS R is different enough that I’m having trouble reconciling that against the competitors. It doesn’t have the resolution of the A7R or Z7, but that not necessarily a bad thing when it comes to the demands it places on both lenses and photographers. The frame rate isn’t as high as the A7 or Z6, but the the resolution is better and loads of FPS, at least in my experience, isn’t always better.
Then there’s the UI stuff that is completely different from the competition. Things like the EF mount adapter with a built in filter slot compatible with polarizers and variable ND filters, or the addition of a control ring to the lenses (and mount adapters). Or for that matter, someone finally building a mirrorless camera that closes the shutter to protect the sensor when the lens is removed.