A Long Ramble on Mirrorless Cameras
It looks like 2018 may just end up being the year of the mirrorless camera. Not that everybody didn’t already have a mirrorless solution of some sort. However, now both Canon and Nikon are expected to announce full frame, serious, pro-grade, mirrorless platforms before the end of the year.
It would stand to reason, presumably, that this should be an exciting new stage in photography, right?
Well, I’m not convinced, or rather I’m no longer convinced.
If I was writing this a year ago, in late 2017, I would have been singing a completely different song. Back then, if Canon were to release a mirrorless version of the 5D mark IV, I would have jumped pretty much instantly. After having spent a week shooting wildlife in Yellowstone, I’m definitely not on that train anymore.
Don’t get me wrong here, it’s not like I don’t recognize the pros of a mirrorless camera. Nor am I just a luddite that’s afraid of change. The problem I have is that, when push comes to shove, a mirrorless camera isn’t always better than an DSLR for my uses.
Mounts, Lenses, and Power…
The big advantage, at least as far as lens design is concerned, of a mirrorless camera is that there’s no mirror in there to push the lens away from the sensor.
The advantage of a shorter register is that more of the shorter focal lengths can be covered without using a retrofocus design. Moreover, where retrofocus designs are needed they’ll won’t need to be as strongly retrofocus.
Admittedly, this is a bit of a double edge sword. Retrofocus lenses produce light that’s more perpendicular to the sensor’s surface. This reduces corner vignetting, as the supporting structures (e.g., micro lenses, color filters, and read out electronics) of digital seniors shad the photo receptor from light that’s approaching at oblique angles.
On the other hand, designing and optimizing retrofocus lenses is more complex. Moreover, the lenses will need more elements, which increases their materials cost and weight. For example, comparing Canon’s EF 24mm f/1.4L USM II to Leica’s Summilux-M 24mm f/1.4 ASPH, the Canon lens has 13 elements in 10 groups to the Leica’s 10 elements in 8 groups. The difference in weight is 650 g for the Canon to 500 g for the Leica[1].
Fortunately, a lot of the vignetting problems of a non-retrofocus lens can be overcome with a bit of technological progress. Back side illuminated sensors help a lot on their own, as does offsetting the micro lenses around the periphery. Finally, there’s always software corrections, which while detrimental to dynamic range and low noise performance are super easy to apply.
In broad strokes, that’s the argument for a shallower register. And there’s a lot going for it. Moreover, as the EOS M cameras (and Sony’s Alpha mirrorless cameras) show you can easily and readily adapt a DSLR lens to a shorter register mirrorless camera with a simple spacer.
That’s great then, so what’s my beef?
Strictly speaking, I don’t have a problem with a short register mirrorless camera. At least not one that’s objectively defensible.
The problem for me comes down to the system as a whole as I would end up owning it. One of the fundamental points I keep coming back to is that while mirrorless cameras are great tools, they still have limitations. For me the biggest limitation is the constant use of power just to be able to observe something through the lens.
See, when I’m shooting wildlife, I tend to spend a lot of time using the camera and a big lens as a spotting scope. I’m poised ready to shoot but waiting for the animal or bird to do something. I don’t use binoculars or a spotting scope for this because when something happens, it almost always happens fast and I’ll miss it while putting down the binoculars and picking up the camera.
With a DSLR, doing this takes no power at all. While I’m sitting there watching the perched eagle, the camera isn’t churning away running the sensor, the processor, and a display. In fact, other than pinging the GPS unit every 30 seconds[2], the camera isn’t really drawing any power from the battery until I start shooting.
On the other hand, with a mirrorless camera I have to make a call; either I can run my battery down to use the camera as a spotting scope, or I can use binoculars and probably miss the action I’m waiting for.
Further, the differences in power consumption is actually quite staggering.
Sony’s A7R III uses a 16.4 wH battery and is rated for only 530 shots through the EVF[3]. On the other hand, Canon’s 5D mark IV uses a 14 wH battery and gets a CIPA[4] rated 900 shots per charge with the OVF. Even with a nearly 20% larger battery, the mirrorless Sony gets at best 70%, and at worse only 58%, of frames as the SLR does[5].
As things currently stand, the only real change I see happening in my work is increasing the amount of wildlife I shoot — largely in opposition to the decrease that’s happened over the last few years as I’ve shot more landscapes. Because of this, and the way I find that I use the camera, I’m really not looking to replace my DSLRs with mirrorless anytime soon simply because of the power considerations. Power is fine when you’re in a city, or a studio, but having to either recharge all the time, or carry tons of batteries, in the backcountry isn’t a desirable situation.
The thing is, looking at things in the longer term, mirrorless cameras can never completely reach parity. They can get bigger batteries (though in the case of Sony’s Alpha series, they already have bigger batteries than Canon or Nikon’s DSLRs), and they can use more power efficient processors. But the disparity never goes away, any advancements made in power for mirrorless cameras can be applied to a DSLR. The problem is simply one of use versus non-use.
So let me get back to the problem of the mount.
Because of the inherent limitations of mirrorless cameras, I don’t see myself moving to a pure mirrorless setup any time soon. That ultimately means a mix of mirrorless and DSLR cameras in my kit[6].
Any camera with a shorter register can have lenses with a longer register adapted with a simple spacer. In fact, we already see this in native mounts for at least Canon’s EOS M series. The EF-EOS-M mount adapter has no optics, and is completely passive electrically having just wires to carry the connection from the EOS M to the EF lens attached to it. The cameras detect and talk to the lens using their native protocols and signaling, and everything works exactly as it should.
The problem is going the other way. It’s much more difficult to adapt a short register lens to a long register camera. It can be done, but the adapter needs to have optics, and generally the adapter will act as a teleconverter.
In a mixed camera system, what I envision having for at least the foreseeable future, mixing and matching lenses becomes a potential problem. It becomes even more of a problem when viewed through the lens of weight and travel. Which bodies do I take? Which lenses? If my best quality ultra-wide zoom is the short-register mirrorless lens, then I have to take the camera too.
With everything on a DSLR mount, all of that goes away. I don’t have to compromise on either not having some lens or having to burn a lot of power because I only want to take one camera instead of two, and so on.
Body Size…
Another advantage that’s often cited as coming along with the shorter register of a mirrorless camera is the ability to make the camera thinner and smaller because of that. After all, if you don’t need 25mm of mirror depth in the body, you can shave off most of that 25 mm (1 inch) from the overall depth of the camera and make the whole package smaller.
The problem is I’ve not yet found a smaller camera that I really find ergonomically sound. And bear in mind, I started out with a Canon Rebel XTi, which is basically about the size of a Sony A7R III or A9.
At this point, about the smallest body that I’d really be interested in using is something in the range of the 6D mark II. Though honestly, I’m very happy with the size of the 5D mark IV.
This is of course, entirely subjective. This is what fits in my hand comfortably and what I find balances well with the kinds of glass I shoot with. In fact, I have battery grips on my cameras in part because I find that the increased size makes them more comfortable.
I don’t want a camera the size of Sony’s A7R III or A9. I’ve already been there with the Rebel XTi, and I don’t want to go back. Moreover, when you look at the kinds of compromises that have to be made in terms of controls, and their accessibility, on a smaller camera that’s also not an ideal I find worth shooting for.
While I don’t begrudge anyone else for liking a smaller camera, I’m very much concerned that the whole gamut of pro grade mirrorless cameras are going to be made small because they can be.
And that’s ignoring the decreased usability that comes with trying to fit buttons in to usable places on a smaller camera.
Weight Savings…
The other big arguments that comes with the smaller chassis is that everything will be so much lighter too. Even though this basically is wishful thinking when you get right down to it.
At face value a smaller camera should be lighter than a bigger one. It’s smaller, there’s less material (e.g., metal forming the chassis) ergo it should be lighter. The problem here is two fold.
The first problem is that the savings isn’t as clear as just looking at some stats would make it seem to be.
When you look at the weight disparity between say the Sony A9 and the Canon 5D mark IV, it’s only 127 g (4.5 oz) — and the 5D mark IV only 143 g (5 oz) heavier than the A7R III. That’s really not a whole lot. To put some perspective on that, it’s slightly less than the weight of a Canon EF 40mm f/2.8 STM lens.
However, the comparison isn’t apples to apples. A ILC doesn’t do anything useful on its own, we need a lens.
This brings us back to the discussion about register and focal lengths, and so forth. The advantage of the shooter register, broadly, was that it allowed wider angle lenses to be made more simply.
However, for longer focal length lenses ones in the “short to medium” telephoto range, like 50mm and 85mm primes or 24-70mm zooms, the distance lost in the register has to be made up by the lens.
Remember, the focal length tells us how far behind the rear nodal point the lens will focus collimated light. For short focal length lenses, the ones that had to be made retrofocus, this would have fallen between the lens mount and the sensor, not on the sensor were it needs to be. However, for lenses in the range of say 50 mm to 85 mm, give or take, the lens on the mount naturally sits about where it needs to be. Reduce the register, and you need to increase the length of the lens to make up for the loss. Some of the weight lost to the thinner camera, is still there but it’s been shifted to the longer lens.
For example consider Sony’s FE 24-70mm f/2.8 GM lens compared to Canon’s EF 24-70mm f/2.8L II USM. While the optical designs aren’t identical, both use 18 elements in 13 groups, neither support in lens IS, and they both have 9 blade electronically controlled rounded apertures.
The Canon lens is 23 mm (0.9 inches) shorter than the Sony, which is only 3 mm shorter than the differences in the two platforms registers. Moreover, the Sony is 81 g heavier than the Canon.
A 5D mark IV with an EF 24-70mm f/2.8L II USM weights 1,605 g (3.54 lbs.) to an A7R III with the FE 24-70mm f/2.8 GM at 1,543 g (3.4 lbs.).
Sure this isn’t indicative of all cases, and of course all things aren’t equal, Canon’s and Sony’s designs are the same ether optically or in terms of manufacture. But it’s not hard to run into a lot of common scenarios where the weight saved on the Sony body is partially picked up by the lens.
Where you see savings across the board with the short register Sony’s is where the short register allowed for simpler designs. Sony’s FE 16-35 f/2.8 weights 110 g less than Canon’s, and their FE 12-24 f/4 weighs less than half of Canon’s EF 11-24mm f/4[7].
And note, I’m using Sony here as being representative of a small body, light weight, short register full frame mirrorless camera; in no small part because they actually have a product on the market with a reasonably complete line of lenses.
But this just leads us to the larger of the problems with weight; what is and isn’t too heavy isn’t an objectively definable goal. Sure a smaller camera can, and likely will, be lighter than a larger one. However, the value of that to the user is entirely subjective.
I shoot with a lot of large heavy lenses, for me a heavier camera shifts the center of gravity rearwards which makes the whole setup feel better balanced and comfortable. If you only ever shoot with super light weight primes, then balance isn’t as much of an issue and lighter weight all around probably is.
Coexistence…
It seems reasonable, given this ramble, to come to the conclusion that I don’t like mirrorless cameras. Truthfully, I think they’re very interesting and intergang tools. But I’ve been around technology and photography long enough to not really get caught up in the idea that there’s a great revolution that’s going to make everything way better. At least not so much anymore.
On the other hand, this is something I see a lot in discussions. It seems to me, reading just about any discussion about cameras, especially mirrorless cameras, an unspoken presumption that mirrorless cameras should supplant DSLRs.
For a long time the argument seemed to be that it’ll happen when the AF tech reaches a point where they can compete with the phase AF systems in an DSLR. At this point, it’s pretty clear to me that with tech like Canon’s DualPixel AF, this has largely happened[8]
I go back to many of the points I’ve repeated in this post though. It’s not that mirrorless cameras aren’t good, or capable of being used essentially in palace of a DSLR in most situations. It’s that they have inherent behaviors that by design cannot be overcome, and that in some instances these behaviors are, for me at least, as significant determent to their use.
If I were to phrase this in terms of set theory, a mirrorless camera is a subset of the capabilities of a DLSR. There’s nothing that can be implemented in a mirrorless camera with respect to image processing or capture that can’t be done by a DSLR (in live view mode). Even an EVF can be provided with either a hybrid viewfinder or an add on EVF module. However, a mirrorless camera can never have an TTL OVF added.
Put simply, I don’t see a mirrorless camera as a complete replacement for a DSLR.
Unfortunately, that’s not how the vast majority of people seem to see things. And ultimately, that’s going to push DSLRs into increasingly niche status as the camera makers follow the demands of the market.
So What do I want in a Mirrorless Design?
With all that said, I guess I should get to the question I set out to answer when I started writing this. What do I want to see in a mirrorless camera.
Honestly, the biggest thing I want is a camera that’s not shackled by being made super small and light. I really would love to see Canon release a mirrorless version of EOS 5D.
If I was building it, I would use essentially the same chassis as the 5D mark IV, though I would lighten it where possible. That should be doable, since only the sensor needs to be held in tight alignment with the lens mount, which means you don’t actually need as much strength in the frame internally to keep everything else (OVF, AF sensors, etc.) aligned.
Overall I would expect the weight to be much closer to the weight of an A7R3 or A9. Especially once the mirror, mirror drive, and OVF are removed. Right now the cameras are only a 150 g apart in the worst case, shaving that down to 50 g after the removal of the OVF and mirror shouldn’t be that to difficult.
I’m clearly not enamored with a new lens mount. I can appreciate value of a shallower register, but again my own personal preference is a unified system that’s got both mirrorless and DSLR options.
Fundamentally the problem with the mount, at least as I see it, is splintering the larger ecosystem. I seriously doubt that Canon’s new mirrorless option won’t natively support talking to their EF lenses — even if they use a shorter register, the cameras will still speak EF to EF lenses natively. The problem is when they start releasing mirrorless designs for various lenses that are superior to the SLR versions (either because they can be due to the short register, or because they simply stop relapsing SLR lenses).
Unfortunately, I think the “popular” sentiment is very much the other way here. Looking at the comments regarding the leaks for Nikon’s Z mount mirrorless camera, the response is overwhelmingly along the lines of bashing Nikon for making it so big. What I can’t tell is if this is honest commentary, that is a lot of people really find current DSLRs to big to hold comfortably. Or if people are just caught up in the current consumer electronics craze of making things smaller for the sake of being smaller.
-
Though the Leica lens is more expensive $7500 to the $1500 for the Canon, this is largely do to differences in manufacturing (Lecia lessen are hand crafted) and economies of scale. ↩︎
-
I’ve switched between running the GPS in my 5D4s at 2 minute intervals, 1 minute intervals, and 30 second intervals so far, and have seen no practical difference in battery life across those settings. This is also in GPS mode 2, where the camera polls the GPS while asleep but not when powered off. ↩︎
-
Or 650 shots with the rear LCD. Based on CIPA testing see[4:1]. ↩︎
-
The CIPA standard for battery testing can be found here. And no, I really think it’s a poor test methodology that doesn’t necessarily reflect any actual camera usage. ↩︎ ↩︎
-
A quick table showing the raw CIPA data for the Canon 5D mark IV versus the Sony A7R III.
Camera Battery Capacity CIPA Frames milliwatts / Frame 5D4 (OVF) 14 900 15.5 5D4 (LCD) 14 300 46.7 A7R3 (EVF) 16.4 650 25.2 A7R3 (LCD) 16.4 530 31.0 -
Which may keep me away from mirrorless cameras all together in for foreseeable future, as my current kit consists of 2- 5D mark IVs, and I have to say this is the most brilliant camera gear setup I’ve ever had. I no longer have to care which camera I grab when I go to shoot something, there’s no disparity in capabilities or image quality at all. And this is a fantastic thing in my opinion. ↩︎
-
The comparison of Sony’s FE 12-24 mm f/4 and Canon’s EF 11-24 mm f/4 isn’t nearly as fair as the focal ranges might suggest at first glance. The Canon lens is 8.3% wider than the Sony, and while that doesn’t sound like a lot, at very wide angles it requires substantially larger front elements (also the heaviest part of the lens) to get that kind of coverage. ↩︎
-
Sony’s embedded phase AF is also certainly good as well. However, I have functional issues with an AF system that replaces any appreciable percentage of imaging pixels and just relies on image processing to interpolate over the missing spots. For small and isolated areas (such as hot pixel removal) I’m fine with this, but I’d rather not have this be the staple behavior of my camera. ↩︎
Comments
First:
I started reading your site after searching Google for some information regarding registering AF points. I find your articles entertaining as you approach most topics from a technical standpoint. I, too have a background in computer engineering and enjoyed the intelligent discussion of the EF-M protocol. Articles like that are probably not for everyone, I just wanted to tell you they _are_ for someone. ;)
As for the article:
I also don’t think mirrorless is the solution to all problems as it’s made out to be (mostly by adopters of Sony cameras). I own an EOS M5 for lightweight shooting, and the idea that just leaving the camera idle still consumes battery is always stressing me out a little in the back of my head. I also prefer a bigger camera body (shooting a 5D4+grip and a 1DX2), especially with front-heavy lenses like the Canon 11-24/4 or the Sigma 14/1.8.
So why am I still excited about a Canon mirrorless?
Well, we know Canon isn’t the fastest company to implement new features to market. Aside from DPAF, which was and still is revolutionary, the envelope what cameras can do has mostly been pushed by Sony. This has gone so far that the A9 and A7* III are regarded as some of the best cameras on the market, even though there’s no complete ecosystem for them (missing lenses, no speedlites, no CPS equivalent).
So far Canon could claim their pro DSLRs were targeting a different market, but a pro mirrorless camera will be compared to the Sony offerings 1:1 and Canon will get bad press if the feature or price disparity is too large. For the first time in years, there’s pressure on Canon to really innovate and not hold back on features.
So I’m expecting (hoping) for a little more than a 5D4 without a mirror (even more so two years after the 5D4’s release). Maybe we’ll see “pro” framerates for stills (even the prosumer EOS M5 can do 9fps), which seems plausible since you save time not having to actuate a mirror. Maybe we’ll finally see a usable 4K codec at a sensible crop factor in a Canon. Maybe we’ll see Canon log in the body, especially as they made it available in the 5D4. Maybe we’ll see Zebras or other video features. Because if we don’t, there’s really little reason why people new to the market should consider the mirrorless FF from Canon over Sony.
It’s certainly going to be an interesting reveal. :)
Hi Christian,
First, thanks for sticking around through my, often overly, long posts. Also it’s good to know that my technical content gets read and appreciated by someone. You’ll also probably be interested in the article I’m currently working on, an RFC for a more extensive power use testing protocol.
I keep waffling on the whole mirrorless thing, I’m big into the idea for a while, then I go back to shooting wildlife and I’m very much back on the fence. It seems to keep coming back to what kind of photography I’m most active in at any given moment.
Honestly, I think Canon gets an undeserved bad rap for this because of selective reporting and memory, a focus by the camera media on what can be pumped in 250 words or less not substance and analysis, and people not really understanding how long actual engineering takes.
IMO, the biggest problem with what Sony’s doing is that they’ve turned cameras into another consumer electronics pump and dump game (I’m almost surprised they don’t start an Appleseque Trade Up program). Granted it makes sense for them, they need to iterate rapidly to gain mind and market share (it makes everyone else look unresponsive). I mean if they hadn’t done what they did, they’d probably be out of the ILC camera market now.
However, they’re also created a situation where $3000 cameras are treated more like a smartphone, use for a year or two then kick it into the recycle bin and upgrade. Sure you don’t technically have to replace your Alpha every year, but the pace of change creates the perception.
I think Canon is going to get bad press regardless of what they do. If it’s the same size as an Alpha, they’ll just be copying Sony. If it’s bigger, they’ll be slammed for not getting mirrorless camera are suppose to be small (this is already happening in the comments of Nikon Z rumor articles where it’s been estimated that the Z cameras are moderately bigger than an A9).
The only way I really see Canon winning is by doing what Sony did and redefining what the market is supposed to be. I.e., mirrorless cameras aren’t supposed to be small, unbalanced, cheaply made toys, they’re professional tools and this is how they’re supposed to be done.
At this point, it damn well be more than just a 5D4 without a mirror.
I would have been happy with that 1-2 years ago, but the 5D4 is already pushing toward the end of it’s expected life cycle.
I’ve long been of the opinion that the biggest limiting factor of the 5D series in terms of frame rate, at least for the last 2 generations, has been the mirror travel speed and the power demands of driving a faster mirror system. There’s some support for that with the 1DX2, when used with the older LP-E4 series batteries, the framerate is limited to 12 FPS.
This has already been addressed. The Digic 8 CPUs can do 4K h.264.
I’m curious as to what the real reason for the MJPEG codec is. On one hand, Canon’s video features in the 5D4 are tricked down from their Cine end of things, not up from the consumer end (note too the DCI 4K format not the consumer UHD format). It’s entirely possible that their advisory board of cinematographers said that they preferred the 1DC’s MJPEG output to a lower bitrate h.264 file (rightly or wrongly).
The other potential problem would be that the hardware simply didn’t have the buffer space for 4K processing. While it seems reasonable to think that they would have addressed this, if they’re coming at video from the Cine end not the YouTube end, that may not have been a consideration.
One thing I’ve been wanting to look at is the PSNR and SSIM differences of 500 Mbps MJPEG versus 100 Mbps ALL-I h.264. I can generate the files easily enough, but I’m still looking for a piece of OSS software that can do the PSNR and SSIM comparisons.
If their mirrorless is aimed at pros, then it almost certainly will have CLOG either from the factory or as a upgraded option.
If your focus is video, this stuff is great. Then again, if I was focused on shooting video, I would own a C200 not DSLR or mirrorless. IME, the ergonomics of a DSLR shaped camera are poor for shooting video.
That depends. Aside from the frame rate, almost everything you’ve listed here are video features. As a still photographer (and a dinosaur who still prefers reading to watching videos for technical content) none of that stuff really matters to me.
I’ve shot 4K clips on my 5D4. The crop (which is just a 1:1 pixel window) doesn’t bother me, and for wildlife is almost helpful. The huge files are kind of annoying, but I can say that about the h.264 files at 1080p and from my GoPro — the reality I see is that working with video just uses gobs of diskspace regardless of format.
Of course, part of the problem is that still photography has been a largely solved problem for 30 years, there’s not much room for anything revolutionary. Even then, there’s a lot of room for subjectivity to creep in as to how valuable further refinement is.
E.g., the A9’s 20 FPS is just not a plus to me. To start with I’m often storage limited anyway, 7 days in the filed will frequently see me shooting over 250GB of stills at 7FPS. Doing it ~3x faster just runs me out of space quicker. Plus there’s the editing problem; I’ve long realized that it’s very easy to get overwhelmed by too many nearly identical images and that really makes editing difficult.
Anyway, thanks for the well reasoned reply. I honestly didn’t think anybody would care about this piece, I just had to get it out of my head.
Just a quick follow-up: I do photo as well as video, ergonomics are kinda the same once you mount the camera on a gimbal so I’m not too concerned “where my video comes from”. I briefly considered a C200, but that camera has it’s own codec size/quality problems…there’s just no middle ground with Canon and video. ;)
The reason I mostly listed video features is as you said, photography is kind of a solved problem. You can maybe still do “faster”, a little better sensor quality, smaller form factor (not everybody wants that) or more niche features. I think mirrorless has little advantages to offer for stills photographers, except for perfectly calibrated AF since you’re measuring focus at the image plane directly. So I guess video features is where it has to be…
The appeal for me: The 1DX2 is a perfect photo camera and just a near-perfect video camera (4K too large, media expensive with currently around 400€/h and playing “disc jockey” a lot). Some video upgrades could swing me towards a smaller mirrorless body as a second video camera, which could then also do photos in a pinch. I doubt though that a mirrorless could completely replace either of my two main cameras at this point.
I agree with your sentiment that Sony is turning cameras into some commodity like a phone. But comparing the original A7 to the III you can see how far it got them. I know some people like to moan about Canon being slow, and sometimes the arguments are ridiculous. But there’s a lot of evidence of Canon holding back not to cannibalize their upper tier…which is their right, but also incredibly infuriating.
Also, I wouldn’t mind the odd feature parity upgrade every once in a while. Something that surprised me: the 5D4 can do servo-AF in Liveview…did you know the 1DX2 can not? That baffles me to no end. It’s just half a year older, it’s running the same sensor and Canon obviously has the software written. So why on earth can’t they push that as an upgrade for their flagship camera? And why does only the 5D4 get Clog? Why do I always have to decide whether I want maximum buffer depth or redundancy…can’t have both, the 5D3, 5D4 and 1DX2 each have one “fast” and one “slow” memory slot. As long as Canon needs to protect their “legacy” upper tier, they’re only going half speed. I’m not sure they can get away with that in the mirrorless market.
The 5D mark IV also doesn’t shoot 16 FPS in live view either.
I play around with embedded systems (Arduino based projects) occasionally for fun, and real-time systems are challenging. You can easily run into problems where you can’t process something fast enough to get out of your own way. Granted I have a lot less compute power than a camera has, but I’m also doing much simpler things.
To have a functional AF system there’s a lot of stuff that has to happen and a lot of that takes time (this might be an interesting article, just to look at form the theoretical engineering perspective). You probably need more than just a frame out of the AF sensor to make predictive measurements. You need to drive the lens motors. And ultimately you need all that done in the window before the next frame needs to be fired off to meet your frame rate targets.
Not offering it at all, or disabling it, is very much in line with Canon’s long standing position that if they can’t make something work all the time and reliably they don’t ship it.
I mean I guess instead we could have another nice confusing table of when things will do what. E.g., “when shooting in live view if servo focus is enabled, frame rates will be limited to X,” or, “when shooting in live view if continuous high is enabled, servo AF will be disabled.”
I don’t know if I like that as a situation, it seems far to prone to creating a situation where in the heat of the moment the camera isn’t working “right” and you can’t figure out why and you’re wasting precious metal resources trying to figure out what’s going on. I’ve been burned in situations in the past, where the camera claimed I could do something but it was outside the camera’s performance envelope and didn’t work right.
Presumably because not enough 1DX2 owners complained/asked for it. The 5D4 didn’t get CLog at release, it was a 6 months later thing, and it apparently wasn’t something that was being worked on prior to the camera’s release.
I’m wishing, because I know it will probably never happen, that with all the storage formats moving to NVMe over PCIe, that we could get user interchangeable slots in the future since everything would just be PCIe from the camera’s CPU. Want two SD Express slots? Take out 2 screwes, pop out the slots in the camera, plug the new slot module in, and redo the locking screws. Want CFExpress, CF, CFast, whatever? Just put the modules in to meet your needs.
You’re right with the processing power and “everything needs to fit together”, but the 1DX2 _can_ do liveview AF when recording video (with object tracking, even at 4K60, which generates about twice the pixel throughput compared to 20MP at 14fps). I know there’s still time required for the shutter, but I don’t buy that it’s impossible in photo mode.
The 5D4 also end up with different LV framerate (~4fps) depending on if you’re using servo or not, and further Canon has put a table in the 1DX2 manual detailing the effect of flicker detection on shooting speed…depending on if you’re living in a 50Hz or 60Hz country! So I’m not sure “we’d need another confusing table” is a good enough reason, because Canon does it for other cameras and also does tables for way freakier stuff than this. :)
What I find is not “right” is that I can track subjects in the viewfinder or in video mode, but not in LV stills mode. I find this more confusing than any possible drop in framerate. :)
Beats me…
But this is really pushing me towards spending money on test equipment now, and I don’t know if that’s a good thing or a bad thing.