Nikon 4.1 Firmware - Initial Field Report

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).

Thank you for linking, I missed it.
I feel you should fwd this to Nikon. What’s the point of using a box if it’s ignoring it? Maybe their NEW philosophy is to use the box only for initial focus, then its free to use the entire frame. Just like the Canon EVF eye system.
I agree 100% - and they know :)
 
One of the most interesting threads I’ve read in a while. I can’t wait for the Z8 to get the Bird.
Its advantages are clear (at least for my use cases) but understanding its limitations will take time. I expect Nikon to improve and refine it over time, as well. It will likely perform substantially better when it’s available for Z8.
 
I agree 100% - and they know :)
iPhone photo of back of camera. I did not observe it jumping anywhere. Of course people need to do real life testing to know for sure.

1696526626678.jpeg
You can only see EXIF info for this image if you are logged in.
 
iPhone photo of back of camera. I did not observe it jumping anywhere. Of course people need to do real life testing to know for sure.

View attachment 71044
I used wide area yesterday (admittedly, I didn't see much, so not ideal testing) but I didn't see it make huge jumps, it tended to stay under/near the box, but it seems like the larger the box is the less likely it is to jump. Of course, more testing needed, etc. I'm hoping to get out this weekend and test it more.
 
It appears there is some room for algorithm improvement though most users would likely prefer eye detect to extend beyond the initial focal point especially for moving subjects. Perhaps a future FW update could provide a menu option to constrain eye detect to the focus box or allow it to use the entire frame?
 
iPhone photo of back of camera. I did not observe it jumping anywhere. Of course people need to do real life testing to know for sure.
I'm not giving you something to do - but try gradually moving closer with each mode until it achieves focus with the eye highlighted. You might also move to a lower position so it looks like a duck - this angle is essentially an oval. :)

With most of these issues, it's not just a yes/no question but more a matter of determining how the function works reliably with a given test.
 
I'm not giving you something to do - but try gradually moving closer with each mode until it achieves focus with the eye highlighted. You might also move to a lower position so it looks like a duck - this angle is essentially an oval. :)

With most of these issues, it's not just a yes/no question but more a matter of determining how the function works reliably with a given test.
Eric, I did that earlier in the thread with a 1X1 Box. And here with the WA-L it does eventually snap right to the eye. What I am suggesting is that it hasn't gone anywhere in the frame. It definitely goes outside the box but with some constraint as to how far.
 
iPhone photo of back of camera. I did not observe it jumping anywhere. Of course people need to do real life testing to know for sure.

View attachment 71044
i would be curious if you put the "ducks" in the center of the frame and then try it. when using auto area it seems to "weight" looking for the subject in the center of the frame. basically i always start with centering the subject, then reframe after subject has been acquired. from a mental perspective, i think of this of like "grabbing it". and if it misses, i try "grabbing it" again by releasing, framing, and engaging af again and usually that works.
 
It appears there is some room for algorithm improvement though most users would likely prefer eye detect to extend beyond the initial focal point especially for moving subjects. Perhaps a future FW update could provide a menu option to constrain eye detect to the focus box or allow it to use the entire frame?
LOL, yet another suggestion I sent to Nikon months ago. I wish they would take at least one two of them!!
 
Test it and find out. I'm fairly confident none of us have this answer.

I'm confident that some of it trickled over, but as to how much, that's a question nikon will never answer (nor would anyone else really).
I'm home now and went out yesterday and didn't think about it, just tested the bird (I assume airplane will be good)...I use auto when I'm shooting surfing and usually am able to catch the oddball pelican or gull. <shrug> Thought someone would know...
 
It appears there is some room for algorithm improvement though most users would likely prefer eye detect to extend beyond the initial focal point especially for moving subjects. Perhaps a future FW update could provide a menu option to constrain eye detect to the focus box or allow it to use the entire frame?
Yeah....like Sony has had since the 2019 A9 FW 5.0 when they introduced RTT. Have RTT on and it will start where you tell it but detect the subject and track it all over. Turn RTT off and it will only detect subject if in the selected AF area. On top of that you can toggle subject detect on/off so that it doesn't look for eyes regardless of RTT on/off.
 
i would be curious if you put the "ducks" in the center of the frame and then try it. when using auto area it seems to "weight" looking for the subject in the center of the frame. basically i always start with centering the subject, then reframe after subject has been acquired. from a mental perspective, i think of this of like "grabbing it". and if it misses, i try "grabbing it" again by releasing, framing, and engaging af again and usually that works.
If I put the ducks in the center and the WA-L box is on the ducks it will go to the eye. Do you want the ducks in the center of the frame but the box off the ducks?
 
Thanks for a great video. I upgraded Z9 this morning to 4.10 and went out to a local park to shoot Norther Harrier. Yesterday I also took photos of Harriers in flight with 4.01, very much around the same time of the day. I usually stand in one place as the harriers fly around me, low and in front of trees and shrubs, sometimes close and more often far away. Both days conditions and EXIF data are very much the same. In terms of keepers rate 4.10 is slightly better over 4.01, 79% as opposed to 77%. Is it a huge difference? Probably not, but it needs more testing. However, the nirvana I was looking for is not there yet. Here are two consecutive frames of the harrier from today as it flew in. I was using 3D tracking for this series. As soon as the bird came in front of a busy background, the focus jumped to the background and stayed there. May be the next version will do better.View attachment 70983View attachment 70984
I suggest auto area AF for this. 3d can be unpredictable and 4.1 has not changed 3d in this regard.
 
If I put the ducks in the center and the WA-L box is on the ducks it will go to the eye. Do you want the ducks in the center of the frame but the box off the ducks?
right. put the desired subject in the center of the frame, it highlights the subject, engage af, keep engaging af and reframe. i'm guessing it'll keep the lock. i do this with running dogs as a normal process.
 
Last edited:
I've tested the new fw today... i feel that with Wide, it finds the eye if it is inside the box, or near it (1 focus point?). if it goes further it loses it, even if the box is in the body of the bird.
With 3d, finds the eye if the box is in any part of the bird.
In automatic, in every part of the screen.
It seems to be quicker to find the eye, and stickier.
 
I've tested the new fw today... i feel that with Wide, it finds the eye if it is inside the box, or near it (1 focus point?). if it goes further it loses it, even if the box is in the body of the bird.
With 3d, finds the eye if the box is in any part of the bird.
In automatic, in every part of the screen.
It seems to be quicker to find the eye, and stickier.
That sounds like it's catching up to the People Eye AF. In the past, animals and birds seemed to lag a little behind people.

My understanding here is this is a matter of training the AI - not coding specific guidelines or standard. Training is a big deal - it's iterative and takes a lot of time. The good news is for the most part it is not hardware limited.
 
What AF modes and settings were you using?
Sorry I was out again all day today. I am getting far better results with Auto Area AF mode with the new bird specific subject detection mode. I used 3D for about an hour today and it was certainly not as intelligent and sticky as Auto Area AF mode.

I am curious if Steve was using 3D AF mode or Auto Area? The Auto Area is virtually instant in recognition/acquisition and just doesn't come off the bird no matter the background and ever a bit further distances then Animal Sub detection. To be clear, 3D is much better using Bird compared to using Animal, but I think it is lagging a tad behind. I think Nikon will need to do another FW update to tweak 3D for the Bird sub detection mode.

@Steve what AF mode did you use for your preview testing for the video? 3D or Auto Area and if 3D, did you at all try Auto Area AF mode?
 
right. put the desired subject in the center of the frame, it highlights the subject, engage af, keep engaging af and reframe. i'm guessing it'll keep the lock. i do this with running dogs as a normal process.
And as @AntonioM said. Keeps the lock if you are IN the box. You go out of the box and it drops. Thus following your target is essential unless you are on 3D.

two more ducks - 1.jpeg
You can only see EXIF info for this image if you are logged in.
two more ducks - 2.jpeg
You can only see EXIF info for this image if you are logged in.
 
Back
Top