Interesting AF comparison A1 vs Z9 vs R3

dibs2010 · Jan 11, 2022

Jared has posted a really good AF performance comparison for the flagship cameras from the 3 brands:

Nikon Z9 vs SONY a1 vs Canon EOS R3 EYE AF REVIEW (2022 Edition) - YouTube

Watch the clip within the video from 3:37sec mark at 0.25 speed in YouTube to see the difference in performance of the 3 top cameras.

To my eyes, A1 looks significantly more stickier than the other two. Z9 seems to be missing the face many times. However, Z9 also seems to be finding eyes at greater distances than A1 which seem to only find the face (and not eyes) when the subject is far away.

I have A1 and my experience has been similar - A1 needs subjects to be somewhat close for human eye detection to trigger - otherwise finds faces only.

Good to see Nikon finally coming in the same ballpark as the other two brand though - competition can only mean better things in the future.

--
-dibs2010
http://flickr.com/dbjunction

Tim S · Jan 11, 2022

If the face is far away then surely depth of field will take care of eye focus.

pollup · Jan 11, 2022

Note that the real test is to look at the actual photos and not the focus boxes. However sometimes it's clear when something is OOF.

From this clip, I would say that Canon and Sony are neck and neck, and Z9 behind.

For example around that time, the Nikon refused to focus on anything other than the background.

87658875fc6241ce8431acede5742dba.jpg.png

dibs2010 · Jan 11, 2022

pollup said:
Note that the real test is to look at the actual photos and not the focus boxes. However sometimes it's clear when something is OOF.

From this clip, I would say that Canon and Sony are neck and neck, and Z9 behind.

For example around that time, the Nikon refused to focus on anything other than the background.

That's true but like you said, some clips clearly show out of focus. If you view this at 0.25 speed, there are tons of instances where both Z9 and R3 miss focus for sequence of shots at stretch.... A1 kept on with him througout except towards to the end where there is a sequence where both Z9 and R3 are in focus and A1 completely misses for a sequence of shots.

--
-dibs2010

dbjunction

Explore dbjunction’s 1,585 photos on Flickr!

flickr.com

dibs2010 · Jan 11, 2022

Tim S said:
If the face is far away then surely depth of field will take care of eye focus.

Yeah... but still

Dspider · Jan 11, 2022

dibs2010 said:
Tim S said:

If the face is far away then surely depth of field will take care of eye focus.

Click to expand...

Yeah... but still

my 2010 pentax point and shoot can detect eyes as well, doesnt mean it has eye af though

eye detect/af is only usefull if your shooting a head and shoulder portraits, even then it focuses on the eyebrow LOL.

Ds

daran · Jan 11, 2022

pollup said:
Note that the real test is to look at the actual photos and not the focus boxes.

Yup. Jared has linked downloadable JPEGs for one sequence. The Z9/Z50f1.2 combo shows a very poor hit rate. I'd blame the stepper motors of the Z50f1.2 rather than the Z9, though.

No account anymore · Jan 11, 2022

daran said:
pollup said:

Note that the real test is to look at the actual photos and not the focus boxes.

Click to expand...

Yup. Jared has linked downloadable JPEGs for one sequence. The Z9/Z50f1.2 combo shows a very poor hit rate. I'd blame the stepper motors of the Z50f1.2 rather than the Z9, though.

Hard to test the Z9 with a non-existing lens ...

pollup · Jan 11, 2022

daran said:
pollup said:

Note that the real test is to look at the actual photos and not the focus boxes.

Click to expand...

Yup. Jared has linked downloadable JPEGs for one sequence. The Z9/Z50f1.2 combo shows a very poor hit rate. I'd blame the stepper motors of the Z50f1.2 rather than the Z9, though.

Oh, big respect to him for doing that then! Some of the youtubers only show the boxes and don't even look at the photos let alone share them.

daran · Jan 12, 2022

Magnar W said:
daran said:

pollup said:

Note that the real test is to look at the actual photos and not the focus boxes.

Click to expand...

Yup. Jared has linked downloadable JPEGs for one sequence. The Z9/Z50f1.2 combo shows a very poor hit rate. I'd blame the stepper motors of the Z50f1.2 rather than the Z9, though.

Click to expand...

Hard to test the Z9 with a non-existing lens ...

What are you talking about? "Lens Model: NIKKOR Z 50mm f/1.2 S"

daran · Jan 12, 2022

pollup said:
daran said:

pollup said:

Note that the real test is to look at the actual photos and not the focus boxes.

Click to expand...

Yup. Jared has linked downloadable JPEGs for one sequence. The Z9/Z50f1.2 combo shows a very poor hit rate. I'd blame the stepper motors of the Z50f1.2 rather than the Z9, though.

Click to expand...

Oh, big respect to him for doing that then!

Jup.

pollup said:
Some of the youtubers only show the boxes and don't even look at the photos let alone share them.

Which is almost completely pointless.

paul cool · Jan 12, 2022

i find the canon af points to distracting the sony i prefer square to small eye af box ,the nikon is quite nice but seems to jumpy and sometimes a little slow to get back on the subject ,all though they all have mistakes the sony grabs jared arm and thinks it is a eye ,would like to see more tests with racking zoom at the same time a real challenge for af .

havoc315 · Jan 12, 2022

Thing is with these types of tests: They show how sticking the AF box is, not necessarily whether there is actually accurate AF.

In other words, there may be times when the box is sticking, but AF actually misses. And times when the EVF AF box hasn't quite caught up to the right spot, but the camera actually is nailing the AF.

The good news -- All 3 of these cameras are nailing precise autofocus a very high percentage of the time, in ways that were impossible just a few years ago.

I think back (not that long) to having to "focus and re-compose" portraits -- which actually meant the eye was never exactly totally focused, you lost centimeters of precision every time you re-composed.

paul cool · Jan 12, 2022

havoc315 said:
Thing is with these types of tests: They show how sticking the AF box is, not necessarily whether there is actually accurate AF.

In other words, there may be times when the box is sticking, but AF actually misses. And times when the EVF AF box hasn't quite caught up to the right spot, but the camera actually is nailing the AF.

The good news -- All 3 of these cameras are nailing precise autofocus a very high percentage of the time, in ways that were impossible just a few years ago.

I think back (not that long) to having to "focus and re-compose" portraits -- which actually meant the eye was never exactly totally focused, you lost centimeters of precision every time you re-composed.

yes and so is the a74 for 3-4k less got to be a sony win either way

Jeff_0000 · Jan 13, 2022

Joe Lynch · Jan 13, 2022

I'll get to see how the A1 focuses for myself over the next week. I pick up an A1 rental and 3 lenses later today. But all I have to compare it to are Olympus bodies and lenses. They all look pretty good to me.

Dspider · Jan 13, 2022

Jeff_0000 said:

the A1 nailed it.

Ds

MILC man · Jan 13, 2022

havoc315 said:
Thing is with these types of tests:

the problem with generalizations like that is that they aren't always accurate, that's why it's called a generalization.

havoc315 said:
They show how sticking the AF box is, not necessarily whether there is actually accurate AF.

true, we see that in the framegrabs below, but the o.p. emphasized a1 stickyness in his post, and there is little question that the a1 wins there by a solid margin, it's obvious by the blurring in the photos.

see the sequence where jared was jumping in and out of the frame, the a1 is the only camera to hold the position of the focus plane, so he's immediately in focus when he jumps back into the frame, ala the first comparison below.

the following comparison where he jumps out of the frame was taken a second later per the timecode, and the z9/r3 both immediately jump focus to the fence in the background.

the r3 doesn't even attempt to follow his travel to the right side of the frame, vs. the other two cameras that have focus boxes on the right edge of the frame.

stickyness can of course be tweaked in firmware.

Dspider · Jan 13, 2022

havoc315 said:
Thing is with these types of tests: They show how sticking the AF box is, not necessarily whether there is actually accurate AF.

In other words, there may be times when the box is sticking, but AF actually misses. And times when the EVF AF box hasn't quite caught up to the right spot, but the camera actually is nailing the AF.

The good news -- All 3 of these cameras are nailing precise autofocus a very high percentage of the time, in ways that were impossible just a few years ago.

no it wasnt, the a6300 got a near perfect score on dpr test for erratic subject towards the camera.

havoc315 said:
I think back (not that long) to having to "focus and re-compose" portraits -- which actually meant the eye was never exactly totally focused, you lost centimeters of precision every time you re-composed.

networker · Jan 13, 2022

Dspider said:
Jeff_0000 said:

Click to expand...

good job

Member said:
the A1 nailed it.

Ds

No it didn't. Almost every picture is front fokused. Mouth and nose are sharp, not the eye - especially at close distance at the end of the video. Maybe use a bigger srceen to view the video ? 4k 32"

Interesting AF comparison A1 vs Z9 vs R3

Leading Member

Leading Member

Senior Member

Leading Member

Leading Member

Dspider

Guest

Senior Member

No account anymore

Guest

Senior Member

Senior Member

Senior Member

Senior Member

Veteran Member

Senior Member

Senior Member

Senior Member

Dspider

Guest

Veteran Member

Dspider

Guest

Well-known member

Similar threads

Keyboard shortcuts