Z9 Eye Detection Expectations - Birds in Flight

[BarkingBeans Coffee]

I am curious about what my expectations should be. I was reading the Autofocus book on Eye Detection and saw that the subject needs to fill the frame. Below is a cropped image. Under the site guidelines of image size it's hard to see but the closer Pelican is tack sharp using Wide Area-Large. (I guess I need get a gallery to link on another site)

However, the camera did not pick up the eye. Is it just too far away as the Pelican didn't fill the frame enough? What should I expect? Any examples would be appreciated.

Since it didn't pick up the eye, I didn't engage 3D tracking with my AF-ON button.

I am new to BIF and Wildlife so trying to refine my expectations and up my skills before a trip to Africa.

Thanks in advance. Happy to answer any questions.

You can only see EXIF info for this image if you are logged in.

You can only see EXIF info for this image if you are logged in.

 

[Steve]

Subject detection tends to favor more "classic" looking birds, so it may struggle with a pelican, particularly one in flight (subject detection is often better at detecting eyes in stationary subjects than it is in moving one). In this case, keeping a Wide AF area on the face is probably the best bet.

 

[BarkingBeans Coffee]

Thank you so maybe I should have engaged 3-D tracking and see what it would have locked onto?

 

[Steve]

Thank you so maybe I should have engaged 3-D tracking and see what it would have locked onto?

3D would likely wander around the body unless you really got it on the eye (the yellow helps since 3D likes color). Still, if it were me, I'd use Wide in this scenario and probably turn subject detection off so it didn't try to go for the body.

 

[StarDust]

It should have picked the eye under such condition. Everything seem to be right.

Not sure why it did not. I never took a picture for pelican.

 

[BarkingBeans Coffee]

3D would likely wander around the body unless you really got it on the eye (the yellow helps since 3D likes color). Still, if it were me, I'd use Wide in this scenario and probably turn subject detection off so it didn't try to go for the body.

I think this is an important intersection of concepts that I haven't seen others discuss and you may want to make an edit in the book about (unless I missed it). The combination of factors here and when turning off Animal eye detect matters. I think I now have to figure out how to free up a button to do so. You or someone else may have suggested one as I recall. Thank you for your insights and for your great book on auto focus.

 

[Steve]

I think this is an important intersection of concepts that I haven't seen others discuss and you may want to make an edit in the book about (unless I missed it). The combination of factors here and when turning off Animal eye detect matters. I think I now have to figure out how to free up a button to do so. You or someone else may have suggested one as I recall. Thank you for your insights and for your great book on auto focus.

Yup, I do discuss it in the books

 

[BarkingBeans Coffee]

Testing of linking of IMGBB images for better display and resolution

 

[Alistair]

"I was reading the Autofocus book on Eye Detection and saw that the subject needs to fill the frame." Really? If you are referring to Steve's book, I cannot recall reading that. It is not my experience at all. I find that it will find the eye when the bird is very small in the frame as long as the eye is clearly presented to the lens.

"the closer Pelican is tack sharp using Wide Area-Large.....However, the camera did not pick up the eye" How do you know, Is the eye not tack sharp?

"Since it didn't pick up the eye, I didn't engage 3D tracking with my AF-ON button." In most circumstances, it is my experience that this is seldom beneficial. Once it has the eye, it has it. No need to switch to 3d. In fact it is often detrimental.

"I am new to BIF and Wildlife so trying to refine my expectations and up my skills before a trip to Africa." That is very wise. Practice is the best way to learn this stuff. As a bonus, the more I practice, the better the Z9 AF system seems to get. Take Steve's book into the field with you as a reference. And give the camera 200-300 milliseconds to analyse the scene. You will know when it is ready to go as the various focus boxes will light up and mobilise. I will try to post some samples a little later when time permits.

 

[EricBowles]

I am curious about what my expectations should be. I was reading the Autofocus book on Eye Detection and saw that the subject needs to fill the frame. Below is a cropped image. Under the site guidelines of image size it's hard to see but the closer Pelican is tack sharp using Wide Area-Large. (I guess I need get a gallery to link on another site)

However, the camera did not pick up the eye. Is it just too far away as the Pelican didn't fill the frame enough? What should I expect? Any examples would be appreciated.

Since it didn't pick up the eye, I didn't engage 3D tracking with my AF-ON button.

I am new to BIF and Wildlife so trying to refine my expectations and up my skills before a trip to Africa.

Thanks in advance. Happy to answer any questions.

You mention that your image was cropped, and the amount of the crop is important. How much was your crop? The subject needs to be relatively large in the frame - or at least large enough to easily identify the subject, head and eye. Said another way, if your subject is far enough away that DOF gives you an in focus subject, do you really care whether it focused on the head or the eye?

I would use Wide Small or Wide Large for that situation. It's a matter of keeping the AF box on the subject - particularly the eye.

When the subject is far away or confused about choice of target, it uses a hierarchy based on scene recognition, contrast in the area selected, identification of the body of the subject, then the head of the subject, and then the eye. At each step of subject identification focus needs to be achieved before it is further refined to a smaller and more specific target.

3D needs the subject to be clearly identified, and then it uses color to follow the subject. If you only have one subject, it will likely find it, but if you have multiple subjects color can be useful.

The display in the EVF is a separate stream of data from actual focus acquisition. It does not need to make a display in the EVF for focus to be achieved.

 

[BarkingBeans Coffee]

You mention that your image was cropped, and the amount of the crop is important. How much was your crop? The subject needs to be relatively large in the frame - or at least large enough to easily identify the subject, head and eye. Said another way, if your subject is far enough away that DOF gives you an in focus subject, do you really care whether it focused on the head or the eye?

I would use Wide Small or Wide Large for that situation. It's a matter of keeping the AF box on the subject - particularly the eye.

When the subject is far away or confused about choice of target, it uses a hierarchy based on scene recognition, contrast in the area selected, identification of the body of the subject, then the head of the subject, and then the eye. At each step of subject identification focus needs to be achieved before it is further refined to a smaller and more specific target.

3D needs the subject to be clearly identified, and then it uses color to follow the subject. If you only have one subject, it will likely find it, but if you have multiple subjects color can be useful.

The display in the EVF is a separate stream of data from actual focus acquisition. It does not need to make a display in the EVF for focus to be achieved.

Eric, both images are above. Cropped and uncropped in the first post. Not a lot of crop.
Thanks for the explanation of the DOF.
I was on Wide Large - and I am aware of using the smallest AF area that works.
What I think is interesting is the hierarchy which this experience caused me to go back and learn about. In this case the for this image and a few others the eye was never selected so the camera stayed on the body or head. I don't recall any smaller focus box lighting up so it may have just stayed on the body.
Your last point is interesting and I guess why one should shoot at the highest frame rate you can.

Thank you for your insights.

 

[BarkingBeans Coffee]

"I was reading the Autofocus book on Eye Detection and saw that the subject needs to fill the frame." Really? If you are referring to Steve's book, I cannot recall reading that. It is not my experience at all. I find that it will find the eye when the bird is very small in the frame as long as the eye is clearly presented to the lens.

From the book-
So, how do you squeeze the most from subject detection, particularly when you’re interested in the camera discovering a face or an eye?
I’ve found that the Z9 is far more likely to zero in on a face or an eye when the subject is at close range and the face or eye is easily recognizable in the viewfinder. The farther your subject is from the camera, the more subject detection will favor the body instead of the face or eye. However, distant subjects are often far enough away that depth of field covers any slight misses due to poor AF placement on the part of the camera.

and another one

The best advice is to ensure your subject fills an adequate portion of the frame and only count on subject detection kicking in when the target is in reasonable focus.

So hence my posting so that I can learn what that means.

"the closer Pelican is tack sharp using Wide Area-Large.....However, the camera did not pick up the eye" How do you know, Is the eye not tack sharp?

The eye is sharp, yet the camera showed only the WA Large frame, it never showed a green box on the head or eye. Maybe the eye is sharp due to the DOF as Eric pointed out. It looks sharp to me.

"Since it didn't pick up the eye, I didn't engage 3D tracking with my AF-ON button." In most circumstances, it is my experience that this is seldom beneficial. Once it has the eye, it has it. No need to switch to 3d. In fact it is often detrimental.

True unless you wish to reframe the shot, and this was a learning expedition for me.

"I am new to BIF and Wildlife so trying to refine my expectations and up my skills before a trip to Africa." That is very wise. Practice is the best way to learn this stuff. As a bonus, the more I practice, the better the Z9 AF system seems to get. Take Steve's book into the field with you as a reference. And give the camera 200-300 milliseconds to analyse the scene. You will know when it is ready to go as the various focus boxes will light up and mobilise. I will try to post some samples a little later when time permits.

Thanks for your suggestions and insights.

 

[TonyF]

Sorry if its been covered already but how many frames in this burst? Its really hard to analyse against a single frame. Were there other frames where the eye was picked up? I would tend to go with Steve's (and anyone else's) opinion on subject recognition being an issue plus its confused by two birds one behind the other. For what its worth it looks like its picked up on the vertical edge on the wing of the second bird which wouldn't surprise me if subject recognition was an issue.

However; I wouldnt get too hung up about eye AF with the Z9. For birds against the sky in good light (so long as it recognises the subject and the eye is visible) then the Z9 is probably the best at detecting the eye and does so at incredible distances.

 

[BarkingBeans Coffee]

Sorry if its been covered already but how many frames in this burst? Its really hard to analyse against a single frame. Were there other frames where the eye was picked up? I would tend to go with Steve's (and anyone else's) opinion on subject recognition being an issue plus its confused by two birds one behind the other. For what its worth it looks like its picked up on the vertical edge on the wing of the second bird which wouldn't surprise me if subject recognition was an issue.

TonyF - This was the first frame of this burst. My question originated as I never saw the green box move from body to face to eye. This has been a helpful discussion for me. Thank you.

However; I wouldnt get too hung up about eye AF with the Z9. For birds against the sky in good light (so long as it recognises the subject and the eye is visible) then the Z9 is probably the best at detecting the eye and does so at incredible distances.

 

[Nimi]

In my experience, the presence or absence of a green square around a feature is not perfectly correlated to that feature being in focus. The square is not what focuses the camera, it's just an indicator, which is imperfect and laggy.

 

[Dave Douglass]

Funny you mention shooting pelicans. I just got back from taking lots of shots of pelicans (brown, not white) with my Z9 and had no problems with it grabbing the eyes with 3D and Animal AF on. It didn't seem to mind the unusual bird shape. Here's an example:

You can only see EXIF info for this image if you are logged in.

 

[BarkingBeans Coffee]

Funny you mention shooting pelicans. I just got back from taking lots of shots of pelicans (brown, not white) with my Z9 and had no problems with it grabbing the eyes with 3D and Animal AF on. It didn't seem to mind the unusual bird shape. Here's an example:

I didn't have 3D on to start. I had Wide Area on. I was waiting for Wide Area to catch the eye then switch to 3D.

 

[Ken Miracle]

@Steve has summed up most of what I do. 90% of what I photograph is birds from Turkey Vultures to Hummingbirds and in a wide range of lighting and habitat mostly for ID use first and then for art second. I am a run and gun birder always on the move in a wide range of habitats, terrain and light and 99% hand held because of that.

On my Z9 I use a Z800PF the majority of the time for my specialty photography. The other lenses I occasionally use for birds are a Z100-400 or a Z 27-200. I am always in AF-C and almost always at 20FPS.

I usually use custom Wide Area 1 at 5x3 on shutter half release to start with on most birds in flight. I have my video record button set to recall shooting hold and the only item on it turns off subject detection as Steve says in some situations it is better to have it turned off. I use the feature a lot less after firmware 2.0 came along and I have used the Z9 more.

On large slow moving BIF like a gliding pelican I may go to single point. I have my FN 1 button set to single point AF. I have my FN 2 set to a larger 13x7 custom wide area 2 and use it on BIF depending on how far away they are the the size of the subject use it on swallows closer in and Cooper's Hawks farther out.

I have A7 focus point persistence set to auto which allows for a smooth handoff from whatever I have my shutter half press set to to 3D etc.

I have my AF-On button set to press to go to 3D and do so often when I have focus from my shutter half press AF Area mode. I will go directly to 3D when I know from experience that the bird, distance, light, back ground etc. will make 3D immediately effective.

Find what works best for you and practice, practice, practice you ability to quickly acquire the bird in the view finder and then pan/track it is important even if 3D is working perfectly on a particular subject.

Mostly have fun and enjoy the moment with the birds.

 

[BarkingBeans Coffee]

@Ken Miracle yes I have been having fun. Especially watching their behaviors which is not something I have spent much time doing. Like watching two Grebe chicks brawl like my Bengal cats, or the feeding behaviors of sandpipers. I appreciate the feedback and will continue to practice.

 

[Ken Miracle]

@Ken Miracle yes I have been having fun. Especially watching their behaviors which is not something I have spent much time doing. Like watching two Grebe chicks brawl like my Bengal cats, or the feeding behaviors of sandpipers. I appreciate the feedback and will continue to practice.

I had a Bengal that was a real annoyance to my wife's very large British Shorthair at times. If he got to annoying she put him in his place. That was a perfect cat mixture in a high vaulted ceiling with loft place we used to own. No place to high for the Bengal and plenty of easy access low territory for the British Shorthair. Now we just have a toothless rescued cat of indeterminate lineage

 

[BarkingBeans Coffee]

@Ken Miracle

High yes indeed; nothing is too high for these cats. On his way to the top of the bakery sign so he can watch his kingdom.
This one holds the house record of three broken champagne glasses in one leap.

You can only see EXIF info for this image if you are logged in.

 

[Calson]

A software engineer described his expertise as knowing where the glass walls were with the coding tools he used. It applies to each camera's autofocus system where it is a trial and error process to learn its strengths and its weaknesses. With my 800mm PF the Z9 has no problems focusing on very small birds but with the 100-400mm lens hummingbirds in open air the Z9 fails completely. Fortunately I am not planning any overseas travel to photograph hummers so not a worry.

The Z9 is a completely different autofocus system and I would expect that the larger the area and amount of detail in the frame the more time it takes for the electronics to process the frame and determine what to focus on and what to ignore. I was watching a video of carpenter bees and the camera had no problem focusing on the head of the bee as the wing motion was off to the side and not between the head and the camera. A bird with a slow wing speed should be an easier target for the Z9 than a fast moving bird. With a flock or group of birds flying in the sky the Group AF on the DSLR cameras would grab focus on the closest bird and there was no getting around that unless one could obtain a smaller view angle to have only one bird in the frame.