Tony Northrup - D810 vs. 5D Mk3

[RLPhoto]

B/c the 5D3 simply won't track across the frame, while the D800 will do so marvelously.

User error.

+1 pretty much. RTFM.

Nope. Not user error.

The 5D3 only has the 'capability' to track across the frame using depth information from the AF sensor, which might work for subjects that don't change depth much (e.g. Birds), but doesn't even remotely work for erratically moving subjects that change significant distance from camera (the case when shooting with wide angle fast primes, for example). In other words: doesn't work AFAIC.

Let me put it this way: if the 5D3 were perfectly capable at this, why would Canon have released iTR in the 1D X? Do any of you understand the idea behind using the metering sensor for subject recognition? Or the entire principle behind Sony SLT?

You know what's better than RTFM? Using the freaking camera. It's quite clear none of you responding have actually compared Nikon's latest 3D AF tracking to Canon's, so the authoritative voices with which you speak are rather comical.

But as someone else stated, perhaps that's exactly what makes CanonRumors so entertaining.

Sounds like your upset for your shortcomings of using the equipment and blaming the camera. The 5D3s AF is superb and there are thousands of users who prove you otherwise.

Please RTFM on how to setup your 5D3.

 

[sarangiman]

Well, the 5D III has a number of settings in it's AF firmware that let you configure how it's AF system responds. Canon cameras are well known, at least have been since the 1D IV and 7D, to track subjects coming right at you. There is also a dedicated AF preset explicitly designed for tracking "erratic" subjects (that's even what it says, it uses that exact word.) I've photographed birds on a few occasions when they were flying right at me, and it has maintained the lock (I think pretty much all of those were 7D...I haven't had the opportunity to try with the 5D III yet.)

Trust me I know this. I've tried all combinations, changing individual settings within certain use-cases.

Here's the thing: I really don't think you're going to get it until you've actually tried 3D tracking on a D800/810 vs 5D3 side-by-side.

I'm not sure why you're having trouble understanding this: the 5D3 literally doesn't have the hardware to accurately follow subjects *around the frame* accurately. Not compared to the competition, anyway. It doesn't have a dedicated RGB sensor (which, btw, is also used for spot-metering linked to AF point anywhere in the frame). It uses depth information from its AF system only, and that just doesn't (and cannot, if you think about if for a minute) work as well as dedicated sensor in practice. The 1D X does, although in my experience it doesn't actually perform as well as, say, the D810. It's certainly worlds above the 5D3, though, for subject tracking.

I don't think you need "3D" metering linked in with the AF unit to achieve this....however Canon does it, they know the boundaries of your subject, where the separation between subject contrast edges are, so they still seem to be able to track subjects that are moving towards you. I also remember reading a couple articles by sports photographers early on after the 1D X and 5D III were released stating how much they loved the 61pt AF system system and it's ability to maintain a lock on say a basket ball or soccer player running down the field towards them.

'However Canon does it' clearly indicates you're not sure what's actually going on.

Furthermore, anecdotes of how well it worked for one use-case doesn't rule out how much better a different system could work for that, or other, use-cases.

If you check the Birds in Flight gallery forum here on CR, you'll find quite a number of bird shots where the bird is moving towards the photographer...not necessarily always "down the barrel", which would be the most extreme case of that...but definitely where the subject is moving towards the photographer.

If you haven't been able to get this to work, then I would first look into how the camera was configured. Because it should most definitely be possible. I was doing it with the rather jittery 7D 19pt AF system years ago...so I know it's possible.

Again, I'm not talking about birds in flight, nor even sports - neither of which stress AF tracking as much as moving babies and running brides while you actively recompose your shot. This essentially entails erratic movement in 3 dimensions. For the last time, I'm specifically talking about both z-depth tracking as well as tracking across the 2D X-Y plane. And specifically with respect to the latter, the 5D3 is significantly behind and/or completely incapable in comparison to a 1D X or any pro-level Nikon.

This shouldn't be difficult to accept. There's a reason it's in the 1D X, and purported to be in the 7D II as well. But if you want to see a great implementation of it - just try a D810. Seriously, just try it. Go to a camera store, and slap some 24/1.4 primes on some bodies. Put a 5D3 in 'auto 65 point selection' mode with AI servo, a 1D X in the same mode but w/ iTR engaged, and a D810 with 3D AF tracking mode engaged. Start on a subject in the center (or whatever focus point you've chosen), then move the camera around wildly (or have the subject dance/run around if you want/can). Be amazed at what how the camera can swiftly move the AF point to stay on the subject. Or not.

I'm clearly not going to get through to you via text, but I guarantee if you pick up a D810 and try it you'll be amazed. Every single person I've showed this to (including Nikon users who just didn't know this mode existed) have looked at me with a 'wow' expression on their faces exclaiming 'I had no idea; this literally changes the way I can shoot and what I can potentially do.'

Actually, this is starting to get strangely reminiscent of the DR debate years ago... where one side just doesn't want to accept that their camera can't possibly not do something another camera can!

 

[sarangiman]

Sounds like your upset for your shortcomings of using the equipment and blaming the camera. The 5D3s AF is superb and there are thousands of users who prove you otherwise.

Please RTFM on how to setup your 5D3.

Please use iTR on a 1D X or 3D AF tracking on a Nikon, then come back and attempt to imply I don't know what I'm talking about.

This is about you not understanding my point/use-case, not about me not understanding the camera I've owned and used and stressed and tested for years.

Clearly you don't even understand how subject tracking on the 5D3 vs the 1D X/Nikons work. So who is it again then that needs to be reading manuals/white papers? Or perhaps just UTFC (using the freaking cameras)?

Not that I should even be engaging with someone who used some of the most illogical arguments to counter the reality of the DR differences between Canon & the D800 years ago.

 

[Dylan777]

B/c the 5D3 simply won't track across the frame, while the D800 will do so marvelously.

What was the subject and shooting condition like? How did set your camera?

 

[sarangiman]

Also, don't misquote me. I didn't say the 5D3 AF system is incapable - my hit rate went up by a staggering amount when I went from the 5D2 to the 5D3. Simply b/c I could actually use the off-center AF points for my shallow DOF photography.

What I'm saying is that proper subject tracking with a dedicated hardware RGB sensor leads to a similarly staggering increase in ability to capture certain moments b/c I don't have to manually select the AF point.

I'm telling you the 5D3's subject tracking doesn't work well enough for me, when compared to the alternatives available (dedicated RGB metering sensor, or Sony's SLT system which lets the imaging sensor do the hard work of subject tracking).

What, exactly, are you arguing RLPhoto? That there's no point to having a dedicated sensor for tracking that can take into account more information than just depth + whether or not an object at a certain depth left one AF point and went to another? That just having the latter imperfect & rather convoluted method of subject tracking is good enough? If so, good enough for who? You?

Certainly not for me, since I'm actually aware of & have used the alternatives extensively (have you?). And not for anyone who values 3D tracking/iTR/SLT systems, nor the guys who designed those systems to begin with.

 

[sarangiman]

B/c the 5D3 simply won't track across the frame, while the D800 will do so marvelously.

What was the subject and shooting condition like? How did set your camera?

I've already clarified these in my previous posts. This is getting pointless. I'll just have to create videos to demonstrate my point.

Still not sure why it's so hard to comprehend the 5D3's lack of dedicated subject tracking via a separate image-like sensor. The same sensor that, e.g., does face-detection weighted metering.

That's a spec, not an opinion.

 

[sarangiman]

Perhaps I need to clarify my initial statement:

The 5D3 might start off OK following a subject, but it quickly gets confused - as it must with so little information - as soon as the subject changes position along 3 axes. How would *you* design an algorithm to track a subject based off only phase information from 61 AF points? And how reliable and tenacious do you think it'd be compared to a system that has ~100,000 pixels of color information to help subject recognition?

That's the whole point of the RGB sensor - to acquire more information to stick to the original subject. And the imaging sensor in Sony SLT designs has even more potential to stick to the subject (whether or not this leads to better actual performance is another story entirely).

Or the imaging sensor in mirrorless ILCs - any of you ever try subject tracking on better, modern mirrorless ILCs? The tracking itself will give almost any DSLR a run for its money - but the reality is the DSLR will often still perform (significantly) better b/c of the dedicated PDAF system (which mirrorless ILCs lack).

 

[dtaylor]

Again, I'm not talking about birds in flight, nor even sports - neither of which stress AF tracking as much as moving babies and running brides while you actively recompose your shot.

Haven't had a problem with either since the 7D. Maybe RTFM? ;D

 

[DominoDude]

Hmm, my thoughts about how a sensor system should behave is not necessarily about acquiring more information, but about acquiring correct and vital information and then draw the right conclusions from them.

I must have been extremely lucky earlier this year when I did a pray and spray on a Black-crowned Night Heron, directly after it took off from its day quarters. A short burst (2-3 shots), followed by long one. In all 21 shots, of which 18 had the entire bird in frame, 12 with decent focus. The birds course of flight was much like the shape of a question mark while passing me by during this. It even managed to fly behind a few branches on a tree between us. And while I did this tracking of the Heron I turned, dipped down to my knees and leaned back.
I can't possible have done that with my ancient 7D with such an outdated AF tracking system, right? We should praise the Heron for its ability to take off and fly at a constant speed, and at a constant distance while making sure it overlapped perfectly with my focus point in the view finder.

 

[sarangiman]

Again, I'm not talking about birds in flight, nor even sports - neither of which stress AF tracking as much as moving babies and running brides while you actively recompose your shot.

Haven't had a problem with either since the 7D. Maybe RTFM? ;D

I didn't say it *couldn't* do either. I said there are other cameras that do it much better.

Perhaps you'd do wise to UTFC (use those freaking cameras).

And anyone saying they've never missed an action shot at 24/1.4 or 35/1.4 or 85/1.2 b/c of focus is clearly lying.

Has anyone here making these idiotic comments actually used iTR, or 3D tracking, or the A77 II Lock-on AF? And then found it to offer no benefit over the 5D3/7D AF tracking in a variety of AF tracking scenarios?

Can either of you answer this simple question?

Just b/c your photography doesn't stress your system, doesn't mean that generalizes to all users. And if any of you claim 100% hit-rate with focus and focus tracking, you're flagrantly lying. No camera can claim that.

 

[sarangiman]

Hmm, my thoughts about how a sensor system should behave is not necessarily about acquiring more information, but about acquiring correct and vital information and then draw the right conclusions from them.

I must have been extremely lucky earlier this year when I did a pray and spray on a Black-crowned Night Heron, directly after it took off from its day quarters. A short burst (2-3 shots), followed by long one. In all 21 shots, of which 18 had the entire bird in frame, 12 with decent focus. The birds course of flight was much like the shape of a question mark while passing me by during this. It even managed to fly behind a few branches on a tree between us. And while I did this tracking of the Heron I turned, dipped down to my knees and leaned back.
I can't possible have done that with my ancient 7D with such an outdated AF tracking system, right? We should praise the Heron for its ability to take off and fly at a constant speed, and at a constant distance while making sure it overlapped perfectly with my focus point in the view finder.

Birds & sports players don't change distance relative to camera as dramatically per unit time as do close-up subjects shot with wide-angle (fast prime) lenses.

You're still not getting it.

Let me ask you this one simple question: have you tried iTR on a 1D X or 3D tracking on a Nikon D4s/D800?

Do you understand that these cameras have tech in them aimed at solving a particular problem with AF tracking, or do you think that RGB sensors & SLT for subject tracking is just a bunch of marketing mumbo jumbo or a conspiracy by all these camera manufacturers?

 

[DominoDude]

Hmm, my thoughts about how a sensor system should behave is not necessarily about acquiring more information, but about acquiring correct and vital information and then draw the right conclusions from them.

I must have been extremely lucky earlier this year when I did a pray and spray on a Black-crowned Night Heron, directly after it took off from its day quarters. A short burst (2-3 shots), followed by long one. In all 21 shots, of which 18 had the entire bird in frame, 12 with decent focus. The birds course of flight was much like the shape of a question mark while passing me by during this. It even managed to fly behind a few branches on a tree between us. And while I did this tracking of the Heron I turned, dipped down to my knees and leaned back.
I can't possible have done that with my ancient 7D with such an outdated AF tracking system, right? We should praise the Heron for its ability to take off and fly at a constant speed, and at a constant distance while making sure it overlapped perfectly with my focus point in the view finder.

Birds & sports players don't change distance relative to camera as dramatically per unit time as do close-up subjects shot with wide-angle (fast prime) lenses.

You're still not getting it.

Let me ask you this one simple question: have you tried iTR on a 1D X or 3D tracking on a Nikon D4s/D800?

Do you understand that these cameras have tech in them aimed at solving a particular problem with AF tracking, or do you think that RGB sensors & SLT for subject tracking is just a bunch of marketing mumbo jumbo or a conspiracy by all these camera manufacturers?

You're probably right, I'm not getting it. But at least I got the shots. I must have been lucky. 12 times out of 21. A bird that starts out flying to the right, and ends flying to the left, doesn't have to have a great change of velocity nor distance. And we certainly don't have to dwelve into what an obstacle between me and the target does to the apparent speed for the sensory system.

 

[sarangiman]

Hmm, my thoughts about how a sensor system should behave is not necessarily about acquiring more information, but about acquiring correct and vital information and then draw the right conclusions from them.

I must have been extremely lucky earlier this year when I did a pray and spray on a Black-crowned Night Heron, directly after it took off from its day quarters. A short burst (2-3 shots), followed by long one. In all 21 shots, of which 18 had the entire bird in frame, 12 with decent focus. The birds course of flight was much like the shape of a question mark while passing me by during this. It even managed to fly behind a few branches on a tree between us. And while I did this tracking of the Heron I turned, dipped down to my knees and leaned back.
I can't possible have done that with my ancient 7D with such an outdated AF tracking system, right? We should praise the Heron for its ability to take off and fly at a constant speed, and at a constant distance while making sure it overlapped perfectly with my focus point in the view finder.

Birds & sports players don't change distance relative to camera as dramatically per unit time as do close-up subjects shot with wide-angle (fast prime) lenses.

You're still not getting it.

Let me ask you this one simple question: have you tried iTR on a 1D X or 3D tracking on a Nikon D4s/D800?

Do you understand that these cameras have tech in them aimed at solving a particular problem with AF tracking, or do you think that RGB sensors & SLT for subject tracking is just a bunch of marketing mumbo jumbo or a conspiracy by all these camera manufacturers?

You're probably right, I'm not getting it. But at least I got the shots. I must have been lucky. 12 times out of 21. A bird that starts out flying to the right, and ends flying to the left, doesn't have to have a great change of velocity nor distance. And we certainly don't have to dwelve into what an obstacle between me and the target does to the apparent speed for the sensory system.

Your evasion of my one simple question speaks volumes as to whether or not you're qualified to comment on what advantages all the other systems I mentioned might offer. So thank you for that.

Furthermore, no one would argue that 12 out of 21 is a perfect AF hit rate. Those sort of odds don't always get you the decisive moment. An AF system that helps increase those odds is a noble goal.

You're not even interested in the idea of AF systems that perform better than your 7D? You don't think they exist? What's your point? That the 7D is 'good enough'? At which point I will once again ask: 'Good enough for who? Everyone'?

 

[sarangiman]

What all your responses collectively are trying to say are that whatever system you're using is good enough. Be it a 50% hit rate or what have you.

And none of you have indicated you've even tried the tech in the 1D X, the Nikons, or in Sony SLT (Lock-on AF with the A77 II).

Yet I'm the one who should read a manual, despite my actually trying all the above, and concluding the metering/imaging sensors show demonstrable gains in AF tracking performance compared to what the 5D3 offers. In other words, they show - SURPRISE! - benefits in the exact areas they were designed to benefit.

No way that's confirmation bias... nope, no way at all.

 

[TeT]

I would be curious to see what setting he was using on the pics to illustrate noise. Not buying that one on face value.

I did not like the 5 minute money grab in the middle, but I get it....

and I dont think they showed enough side by side pics, and they needed the settings illustrated on the pics.

when comparing pro bodies, you need to put pro info in your comparison , not a bunch of "blew me away"

Lastly this post was originally submitted by a guest, nice successful TROLL.

 

[dtaylor]

Perhaps you'd do wise to UTFC (use those freaking cameras)...Just b/c your photography doesn't stress your system, doesn't mean that generalizes to all users.

Look at who is a special little snowflake, the only one to ever use these cameras or stress his system, with babies and brides no less :

Canon doesn't dominate pro sports with cameras that can't track. But getting the AF settings right for the subject is critical. Now which is more likely? That you didn't set up your camera correctly? Or that Canon's AF is the worst of all?

Hmmm...

 

[sarangiman]

Perhaps you'd do wise to UTFC (use those freaking cameras)...Just b/c your photography doesn't stress your system, doesn't mean that generalizes to all users.

Look at who is a special little snowflake, the only one to ever use these cameras or stress his system, with babies and brides no less :

Canon doesn't dominate pro sports with cameras that can't track. But getting the AF settings right for the subject is critical. Now which is more likely? That you didn't set up your camera correctly? Or that Canon's AF is the worst of all?

Hmmm...

And thank you as well for not answering my simple question of whether or not you've ever used subject tracking using a dedicated sensor.

To (re-)answer your question: I set my camera to AI servo, then to the 61-point automatic selection mode which, in AI servo mode, is indicated by 4 brackets and the selected AF point outlined by an outer square. I selected the center AF point, then depressed the shutter button half-way to initiate focus; from that point on, the camera's AF system takes over in selecting the correct AF point to track that initial subject.

There are a number of use-cases and I tried all of them, but particularly focused on the ones dedicated to erratic movements in both the frame as well as depth. I adjusted all the individual settings to various degrees. They varied performance to certain degrees, but in all cases performance was so much worse than the 1D X or any of the pro-level Nikon bodies that it's not really worth discussing.

It was far too easy for the system to get confused and lose the subject, requiring re-initiation of focus using the center point.

Let me put it this way:

What's more likely: (1) my settings - which I've outlined above - were wrong, or that (2) maybe, just maybe, all the work put into using dedicated RGB sensors by Nikon for years, Canon playing catch up by putting it in the 1D X, and Sony dedicating themselves to the SLT system to use the image sensor for AF tracking is actually worth some clear benefit to some types of shooting and that you have no idea of this benefit b/c you clearly haven't picked up any of these cameras and used them in these modes?

In other words, what's more likely:

(1) The 1D X offers no AF advantages over the 5D3 (what you're saying)
(2) You just don't know what those advantages are (what I'm saying)

Hmmm...

 

[sarangiman]

Canon doesn't dominate pro sports with cameras that can't track. But getting the AF settings right for the subject is critical. Now which is more likely? That you didn't set up your camera correctly? Or that Canon's AF is the worst of all?

Hmmm...

And, er, why do you say 'Canon' above? You mean the 'Canon 5D3, 7D, etc.', not, importantly, the '1D X'.

This is an important distinction, b/c my whole point in these posts is that the 1D X* offers something the 5D3 cannot do well.

I know that's so hard to believe, that a camera twice the price is actually better, but... well... I've got news for you.

*Amongst others, but let me leave other brands out since apologists take to flames whenever someone mentions another brand can do something better.

 

[RLPhoto]

Perhaps you'd do wise to UTFC (use those freaking cameras)...Just b/c your photography doesn't stress your system, doesn't mean that generalizes to all users.

Look at who is a special little snowflake, the only one to ever use these cameras or stress his system, with babies and brides no less :

Canon doesn't dominate pro sports with cameras that can't track. But getting the AF settings right for the subject is critical. Now which is more likely? That you didn't set up your camera correctly? Or that Canon's AF is the worst of all?

Hmmm...

+1 He really should RTFM because I tried a d800 for a weekend and it's AF sucked like a vacuum cleaner compared to the 5D3. I don't know about the 810 but it still uses that multicam backbone that was in the d700, which I also used before. Also I remember the G 1.4 primes being a slug compared to my 24L II 50L and 135L.

He probably didn't RTFM and I haven't seen any photos from him either.

 

[sarangiman]

Chill with the attitude man. ??? What's up with that? Once you get all snarky, no one is going to listen. We both know that all too well.

Sorry, these responses from people who clearly haven't compared the systems side-by-side is just intolerable. At least I can engage in a conversation with you... (well, except that one time years ago )

Now, you claim that Canon cameras cannot track in the "Z" axis (which I can only assume means within the depth of the scene, towards or away from the photographer), or track in Z while also tracking in X/Y (vertically and horizontally within the frame).

No no no I don't claim that at all. I would never say the 5D3 isn't good at tracking in the Z-axis; in fact, it's phenomenal at it. You got my statement priorities completely backwards in your statement. I'm talking about the 5D3's ability to automatically move the AF point in the X-Y planes to stick with your initial subject as you recompose or the subject moves quickly & erratically in the X, Y, and Z planes.

I disagree that it's literally impossible to do accurately with Canon. It may not be as good. I don't know exactly how good Nikon's 3D AF unit is, I haven't had enough time with a Nikon camera in specific use case scenarios to fully test it out. I am not denying it could be better.

It's not literally impossible. I never said that; at least I hope I didn't. And if I did, I meant that it's so inaccurate compared to what's available in the 1D X and Nikon pro-bodies that - to me - it's almost incapable in comparison. My sincere apologies if I implied it's literally or completely incapable. But I do believe I've said on many occasions that the 5D3 does have a mode that attempts to do this; it just doesn't do it well.

Please, please do yourself a favor and go to a camera store & plop on a 24/1.4 or 35/1.4 and try 3D tracking on the D810. Initiate focus using the center (or whichever) point on the camera salesman's eye (yes, get up that close to him), and then move the camera around wildly recomposing. The camera will 'stick' to the eye so well you'll wonder if there's some trickery going on whereby the camera is using the accelerometer to figure out how to move the focus point (this is obviously not the case).

It'll be so good that your entire perspective on distance + 2D AF tracking across the frame will completely change, and then you might understand why I feel that the 5D3's 'ability' is so poor next to the competition in this regard that you may as well call it 'incapable' in this respect. And that it literally doesn't have the hardware necessary to track as accurately as what the other systems offer. The 1D X will do this sort of tracking as well, but is more prone to losing the subject in all my tests than the system Nikon's honed for years and years. And that, of course, shouldn't be surprising as it's 1st-generation tech for Canon.

That's not to say the 5D3's AF system is bad. I started this conversation b/c someone made a blanket comment that the 5D3 clearly has a superior AF system to the D800. And for certain types of shooting, that's just demonstrably wrong. Why that's so hard for people here to believe is astounding.

And one of the reasons I'm being rather passionate about this discussion is b/c I kept my 5D3 for years b/c I also believed that there wasn't a camera capable of better AF out there. Part of that belief was being mislead by people making erroneous claims just like the one above about Canon AF undoubtedly being better. You might say I'm just allergic to this sort of misinformation.

It's more complicated than what some would lead you to believe. I pointed out where the 5D3 is better (cross-type AF, high sensitivity wider-baseline AF points in the center) and where it's worse (3D AF/iTR). And boom that was it. B/c of that one thing that I claimed that Nikon can do better (as well as the 1D X, just not quite as well), I was attacked for obviously being wrong, should RTFM, etc. etc. uninformed opinion, etc. etc. BS from people who - even up until now - haven't even once answered my question 'Have you used iTR or 3D AF tracking, ever, to so authoritatively claim I'm wrong?' Or, for that matter, tried subject tracking on a mirrorless ILC (not overall focus performance, just the ability for that on-screen box to move around & stick to your original subject)?

Because obviously there's no way any of these guys might have missed something all these years having shot Canon, right? I mean all of us here know everything, right? RIGHT??

Which is why I say that confirmation bias runs rampant on these forms.

I am specifically disputing your claim that Canon cameras (with the exception of the 1D X, although you allude even it cannot do it, despite iTR metering) "literally doesn't have the hardware to accurately follow subjects *around the frame* accurately."

No, I never said the 1D X can't do it, despite iTR metering. I said it doesn't do it as well as Nikon's system, which should be no surprise as Nikon has iterated this system through many generations over years and years, and this is 1st generation tech for Canon. And by that I mean the 1D X is more prone to losing the subject or getting confused. It's still very good compared to the 5D3, which itself has iterated its algorithms over years and years to use depth information to stick to a subject as it moves across the frame.

But how good can that possibly be? Think about it: the center point of the AF system detects a subject 10ft away, then you recompose, then the camera notices a subject 10ft away is now over the left-most AF point, and meanwhile there's now nothing at 10ft away at the center point. Therefore, the camera decides your subject has moved (or you've recomposed such that the subject is now at) the left-most AF point. But what if your subject moved to 9ft away during this time as well? Well, with some clever algorithms you could analyze all the focus points and see if there was some progression of a subject like this (I've assigned letters to specific focus points for ease of discussion):

[list type=decimal]
[*]Subject in center point (C) 10ft away
[*]Center point C no longer detects anything at 10ft, but the point just to its right (D) has a subject at 9.8ft
[*] Point D no longer detects a subject at 9.8ft, but the point to its right (E) detects a subject at 9.5ft.
[*] Point E no longer detects a subject at 9.5ft, but a point 6 points to the left (F) now detects a subject at 9.2ft.
[/list]

... and so on and so forth.

Are you starting to see how incredibly complex this can get, and how prone to failure this might be if the subject is moving like this in 3-axes and/or the movement is convoluted with you recomposing? Or another subject entering the frame at a similar depth?

Are you starting to see how using an image sensor (Sony SLT, or all mirrorless ILCs really), or a color sensor with some finite resolution to recognize color patterns (enough to detect a face, which we know RGB sensors can do given their face-detection ability) that communicates with the PDAF sensor might have the potential to perform significantly better?

Canon cameras ARE capable of tracking in Z, or tracking in Z while the subject is also moving around in the frame. That's what I'm saying, that's what everyone else is saying, and it's been possible since at least the 7D, since I've tracked rather erratic birds flying directly at me (i.e. in the z-axis) with the 7D.

Funny, b/c that's what I've been saying too. Of course it can track in the Z-axis well; that's the whole point of phase detection! At the risk of sounding like a broken record, the 5D3/7D just can't accurately track that subject in the X-Y plane of your frame as well as a camera with an accessory color/image sensor that has the ability to recognize a subject based on an actual, albeit low-resolution, RGB image.

Now, you have re-qualified you statements to specifically refer to close subjects with wide angle lenses.

I only said that b/c there can be a lot of focus plane shift from recomposing using wide-angle lenses, and subjects can drastically change distance and fall out of the acceptable DOF zone. Combine this with movement across the X-Y plane and you're really, really stressing an AF system that can only track subjects based off of analyzing distance information from 61 (or worse, 19) AF points.

Let me try to explain with some numbers:

Your DOF at 35/1.4 for a subject 1m away is 6.4cm. At 2m it's 26cm. Either way, it's small. Your DOF at 200/2.8 for a football player running towards you from 30m away is nearly 4 meters. Now do my thought experiment above for each of those cases. For one, you have more tolerance for error in tracking b/c your football player is not going to fall out of your DOF zone in a fraction of a second. Second, if others enter the frame near this runner - as is typically the case with sports - even if this subject confuses the AF system, it's quite likely it won't affect the perceived focus (b/c they'll both be in a similar plane within the tolerable DOF).

Because using AF tracking to allow the camera to automatically select the AF point to stick to my moving subjects at 24/1.4 or 35/1.4 or even 85/1.2 was so unreliable in practice - in critical scenarios like weddings where you can't afford to have the camera wildly missing focus during the action - I'd resort to selecting my AF point, and then allowing the camera to track (AI servo) along the Z-depth axis only - which it does/did remarkably well. But that constrained me to keep the subject under the selected AF point during the action, which constrained my composition.

My point is that with iTR and 3D AF, you can often actually trust the camera to use the correct AF point after you've initially identified the subject - which allows you to decouple the composition & subject movement from focus.

I honestly don't think messing around with a camera in a store is going to prove the case either...however renting a D810 would be sufficient to let me explore their 3D AF in totality.

It's actually so obvious that I bet you a nickel a trip to the camera store is all you need to get an initial idea of what I'm talking about.

On the other hand, full tracking of subjects at longer focal lengths, including those moving at you in the z-axis (as has often been seen with basketball and soccer photos...where the player with the ball is moving down the field, dodging oncoming opponents), has been done quite frequently. And before the 1D X...I remember there being a lot of talk about how much that improved with the 1D IV, and I remember reading a couple of articles comparing the 1D IV to Nikon cameras in exactly that use case scenario (I can go digging for them, but that was a number of years ago now...) Neither camera performed poorly...although I remember the reviewers having specific complaints about BOTH AF systems in those use cases.

I KNOW that Canon does gather enough information to perform the necessary z-axis tracking analysis.

Again, you have no argument from me here on this. You seem to have completely misinterpreted what I've been writing!

I don't think a full RGB metering sensor is essential for z-axis tracking, even of erratic subjects. It's probably BETTER, Canon clearly states that iTR enhances the tracking abilities of the 1D X AF system.

Er, which is essentially what I've been trying to say all this time. I really wish people actually read and digested what others right, rather than just seeing 'Nikon' and being like 'SQUIRREL!' and jumping behind their barracks ready to assume the defensive apologist attitude and the offensive RTFM/user-error/you-can't-possibly-be-right-b/c-that'd-mean-I've-been-wrong-all-this-time attitude.

And for the record...I always want better. I am very interested in seeing what Canon does when they link their full sized image sensor into the AF system. That could give them the ability to gather pre-frame information via the iTR, as well as full-frame information during the following exposure via DPAF...THAT has me rather intrigued.

Yes, I firmly believe DPAF might revolutionize AF. For reasons that are too lengthy to go into here/now.