R3 vs A1 for sports autofocus

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

FB101

Well-known member
Nice comparison by Jared Polin.
He toned down the theatrics this time and if you fast forward the 5mn infomercial in the middle it is actually well done.

As I watched his EVF capture and resulting pictures I would agree with his conclusion. The R3 seems a bit more sticky, a bit better at getting to a face or an eye, and both cameras eclipse everything else out there (until the z9 can be compared of course). So slide edge for the R3 in that case and for its ability to keep 30fps in lossless raw.

You can also look at it from a different angle. The A1 delivers 98% (my math, completely unscientific, let’s call it a very high percentage) of the R3 performance in what is the R3’s best use case that it was optimized for, while processing twice the resolution.
So the A1, which in my view is the most versatile camera in existence, delivers almost identical performance as a camera that is highly optimized for sports.

And it’s also a huge accolade to canon that they actually were able to push the envelope beyond the A1 for sports - i am not sure I would have placed that bet 6 months ago.

As a side note, I wonder what would happen if you shot the A1 in DX crop. 21mpx vs canon’s 24 - just from an AF perspective, would processing half the data eliminate the gap.

 
As to the AF boxes being more sticky and tracking face/head more consistently on the R3 vs A1, i think its just a matter of firmware update for A1 to match or exceed the R3. Will be interesting to see how the Z9 performs here as it has a similar AF (Eye, Head and Torso tracking).

To me the biggest take away from all this is that the mirrorless specs must be taken with a grain of salt. A1 and Z9 are reported to perform 120 AF calculations per second whereas the R3 is reported at 'just' 60 calculations. The real world AF performance puts the Canon R3 slightly ahead of A1. Maybe the 120 CPS on A1 and Z9 will have an impact when/ if there are significant AF firmware updates but we will have to wait and watch. Like wise the EVF specifications where the Canon is 5.7M dots vs 9+M dots on the A1 however the R3 EVF seems to perform slightly better for fast moving subjects. I have a hunch EVF performance is one area where the one with the least spec (Z9) will end up being the best.
 
Regarding the EVF recording, since it is a camera feed and not filming the actual EVF, do we know if it is the same as what’s being shown of the EVF? I honestly have not looked into that question but I could easily see that on some cameras it could be a dummied down version.
 
The R3 did look impressive. It seemed like the R3 was a bit more responsive and quicker to acquire. I think both look incredible and I hope the Z9 proves to be at a minimum, nearly as good. This level of tracking will certainly make things easier. Looking forward to being able to use it.
 
Excellent real world AF review by Jarred.

I suspected the R3 to be better than A1 in terms of AF, seems like it is.

However, the AF of A1 is excellent too.

I really hope the Z9 does it better than R3 & A1 as seems to have the most powerful processor in the market.
 
The image shown at the time stamp 14.03 is the magic of AI. I don't think any DSLR could have managed that.

I think most of the top end pro sport bodies (d6, 1dx) could have done that if they were properly set up for that specific case. These bodies have parameters for how quickly they will jump to another subject. Properly set up, they will tolerate subjects being obscured briefly and still stick to the subject.

That said, I do buy that these new mirrorless cameras are going to be able to do it better and with less forethought.

Also, not meaning to harp on it, but i think 16:11 - 16:16 shows where I think mirrorless could have done better but didn't. The guy was briefly obscured, then it didn't pick him up on the other side even though he didn't change shape or size or color (or direction) substantially. Optimally the system should have been able to tell it was the same guy and pick him right back up but he appears to go out of focus and then the system kind of hunts around and picks him up later, apparently as a new subject. On the other hand, there are a lot of similar sized, shaped and colored objects, so obviously that's going to make it harder. Of course, I don't know the focus modes and maybe it wasn't set up to try to track him tenaciously, dunno.
 
Last edited:
blocked shot example with the lowly d500. i'm running group af here so the disc is definitely going to be in my af area and group will pick the closest object, so how fast the camera will pick the disc over the dog is driven by that blocked shot setting*. *Thanks to Steve here for his great videos and ebook on the AF settings ;-)
JN9_2906-Edit-Edit.jpg
You can only see EXIF info for this image if you are logged in.
 
Last edited:
The thing that struck me in Jarod's video is how tremendously both cameras performed. Both of these cameras (and I suspect the Z9 will join them) simply do things that couldn't be done only a few years back. Certainly, the very best had good tracking routines, but nothing across the full frame with the level of camera driven decision making as what we currently have. Great time to photograph wildlife and action.
 
I think most of the top end pro sport bodies (d6, 1dx) could have done that if they were properly set up for that specific case. These bodies have parameters for how quickly they will jump to another subject. Properly set up, they will tolerate subjects being obscured briefly and still stick to the subject.

That said, I do buy that these new mirrorless cameras are going to be able to do it better and with less forethought.

Also, not meaning to harp on it, but i think 16:11 - 16:16 shows where I think mirrorless could have done better but didn't. The guy was briefly obscured, then it didn't pick him up on the other side even though he didn't change shape or size or color (or direction) substantially. Optimally the system should have been able to tell it was the same guy and pick him right back up but he appears to go out of focus and then the system kind of hunts around and picks him up later, apparently as a new subject. On the other hand, there are a lot of similar sized, shaped and colored objects, so obviously that's going to make it harder. Of course, I don't know the focus modes and maybe it wasn't set up to try to track him tenaciously, dunno.


Idk, I have used the stickneess feature adjustment on the D6 & D500 & it would often get distracted by other things. Also, when the stickneess is increased, the initial target acquisition is significantly slower.

DSLRs are not programmed to know that a certain object is an eye etc., It is not looking for an eye shaped thing. Mirrorless cameras are.

But we can only speculate. One needs to do a side by side comparison.
 
The thing that struck me in Jarod's video is how tremendously both cameras performed. Both of these cameras (and I suspect the Z9 will join them) simply do things that couldn't be done only a few years back. Certainly, the very best had good tracking routines, but nothing across the full frame with the level of camera driven decision making as what we currently have. Great time to photograph wildlife and action.


Totally, and all those ranting about how one should improve their skills will be quite happy to use the AI subject assisted AF tracking modes on the new mirrorless bodies.
😄

It is time we worry less about our AF skills & focus more on composition skills as that's what make images special.
 
Last edited:
DSLRs are not programmed to know that a certain object is an eye etc., It is not looking for an eye shaped thing. Mirrorless cameras are.

Really not trying to argue, but some DSLRs, like the D6 do have eye detection. No idea how good it is, but *shrug*. And object tracking is different than object recognition.

Again, I'm not saying DSLRs are better. I'm in the process of ditching my DSLR for a mirrorless. I do think they will transform how we shoot. I totally believe that the mirrorless af have a much better "concept" of what construes an object, and can track it based on it's ability to continue to recognize it as the same object in a much more sophisticated way than the best DSLRs.

But I also feel like there can be a certain amount of pointing to things as evidence that may or may not be evidence of what they are claiming. In the case of Jared's video, it really rubbed me the wrong way because I don't feel it was clear evidence. I think most sport shooters with properly set up DSLRs would have gotten that exact shot and there's lots of reasons.

And the constant talk about "stickyness" belies the fact there are very real limitations (like 16:11 - 16:16). I've noticed similar things with the a1, where it looses the object you'd think it could hold onto, or you can see it gets confused about exactly what is included and not included in that object. I'm totally not complaining or criticizing - this is a great time to be a photographer. Just a call to police ourselves from hyperbole or jumping to conclusions.

Again, sorry, not trying to be argumentative here.
 
Really not trying to argue, but some DSLRs, like the D6 do have eye detection. No idea how good it is, but *shrug*. And object tracking is different than object recognition.

Again, I'm not saying DSLRs are better. I'm in the process of ditching my DSLR for a mirrorless. I do think they will transform how we shoot. I totally believe that the mirrorless af have a much better "concept" of what construes an object, and can track it based on it's ability to continue to recognize it as the same object in a much more sophisticated way than the best DSLRs.

But I also feel like there can be a certain amount of pointing to things as evidence that may or may not be evidence of what they are claiming. In the case of Jared's video, it really rubbed me the wrong way because I don't feel it was clear evidence. I think most sport shooters with properly set up DSLRs would have gotten that exact shot and there's lots of reasons.

And the constant talk about "stickyness" belies the fact there are very real limitations (like 16:11 - 16:16). I've noticed similar things with the a1, where it looses the object you'd think it could hold onto, or you can see it gets confused about exactly what is included and not included in that object. I'm totally not complaining or criticizing - this is a great time to be a photographer. Just a call to police ourselves from hyperbole or jumping to conclusions.

Again, sorry, not trying to be argumentative here.
No I didn't find you argumentative. I was making a general point.
I get your point that one can use top end DSLRs better & that mirrorless AF are not flawless.
 
blocked shot example with the lowly d500. i'm running group af here so the disc is definitely going to be in my af area and group will pick the closest object, so how fast the camera will pick the disc over the dog is driven by that blocked shot setting*. *Thanks to Steve here for his great videos and ebook on the AF settings ;-)
View attachment 27842

Nice capture.
 
And the constant talk about "stickyness" belies the fact there are very real limitations (like 16:11 - 16:16). I've noticed similar things with the a1, where it looses the object you'd think it could hold onto, or you can see it gets confused about exactly what is included and not included in that object. I'm totally not complaining or criticizing - this is a great time to be a photographer. Just a call to police ourselves from hyperbole or jumping to conclusions.

Again, sorry, not trying to be argumentative here.

you are spot on. The more we use those cameras the more we find their weaknesses and they do have some. But I keep going back and forth, A1 / D850 and going to the D850 is getting harder every time.
It a blend of the responsiveness of the A1, the benefits of EVF, what I personally find a more reliable AF and much more comfortable ergonomics (again very personal). The only reasons I grab the d850 anymore is because I dont have many Sony lenses, otherwise I wouldn’t grab it at all.
But to your point, it’s very different from saying the A1 is flawless.
 
blocked shot example with the lowly d500. i'm running group af here so the disc is definitely going to be in my af area and group will pick the closest object, so how fast the camera will pick the disc over the dog is driven by that blocked shot setting*. *Thanks to Steve here for his great videos and ebook on the AF settings ;-)
View attachment 27842
Great shot John!👍👍
 
you are spot on. The more we use those cameras the more we find their weaknesses and they do have some. But I keep going back and forth, A1 / D850 and going to the D850 is getting harder every time.
It a blend of the responsiveness of the A1, the benefits of EVF, what I personally find a more reliable AF and much more comfortable ergonomics (again very personal). The only reasons I grab the d850 anymore is because I dont have many Sony lenses, otherwise I wouldn’t grab it at all.
But to your point, it’s very different from saying the A1 is flawless.
Same here I hardly touched D 850.How ever I had to use today D 850 with 85 mm for my sister's German shepherd indoor dog shoot since I have only 200 600 for the A1.While I got pretty good shots I did miss the Animal Eye AF of A1
 
Yah, I'm in no way suggesting the DSLRs are going to be more desirable than the top end mirrorless. Personally, I'm not even going to keep my DSLR. I just hear a lot of what sounds like misattribution and hyperbole in these Youtube videos and sometimes it rubs me the wrong way.
 
As an aside, I think Jared's EVF footage really shows the challenges some sports shooters face with lots of clutter. I think the chaos seen in his EVF at times really shows has the user interface is going to be a huge part of making a camera successful in these environments.
 
Does anyone else think fro's EVF footage shows a lot of stuttering or dropped frames? also why does the SONY EVF feed look to be a different aspect ratio?
Yes, but my takeaway on it is that that it’s still better then having blackout. I also wonder if any of it is related to being a feed of the EVF. For example, the Z series has a more exagerated blackout in the EVF feed than the actual EVF. I think the aspect ratio is an appearance as with the canon the image is using the entire display area with the settings overlaying the image and the Sony has a smaller image with the settings in the black area below the image.
 
Back
Top