RX10 IV Firmware 2.0 Bird Eye AF

[AlwynS]

The efficacy or not of the new firmware 2.0 release for the RX10 IV has been discussed in some other threads. I have recently installed the new firmware and wanted to find out whether/how it works.

This morning, I went to a local migratory bird park (Reifel) with my good friend Steve (evetsf on this board) from Bellingham, WA . This was a good opportunity to check the Animal Eye AF on birds. And I have come to a pretty much unequivocal conclusion:

With slow moving birds on the ground and on water, VERY close up, Eye AF works as advertised. BIF or birds NOT filling a very significant part of the frame: no way. Things get tricky where the eye is relatively close and the bird moving slowly BUT the eye is hard to distinguish, e.g. black eye in a black head. Which should probably not be a surprise.

When I say close up, the following shows where I found Eye AF typically worked for me. Note that all items are cropped to no smaller than 3840 x 2160 with several being larger than that.





Eye tracking recognised they eye on this Coot and generally even held on when he started turning his head away until they eye was essentially fully obscured.







Similar to the Coot







Generally tracked:



Some unsuccessful examples to follow.

Note that when I used Flexible Spot, once the Eye Focus locked on and started tracking, it immediately lost tracking when the eye went outside the Flexible Spot. Something to bear in mind.

--
Cheers
Alwyn

 

[AlwynS]

The following are examples of the types of situation where I had essentially zero success:

Too far







BIF, too far, too fast?



Eye not readily discernible:



--
Cheers
Alwyn

 

[AlwynS]

The following are examples where Eye AF was erratic:



Eye reasonably large but not prominent/discernible enough



More off than on:



Also more off than on: very close but hard to discern I think.



I have a lot more examples of each type but I just wanted to show a few reasonably representative images

--
Cheers
Alwyn

 

[Rodger1943]

An interesting mix of images, which all look good, whether eye AF was working, or not. I have installed Firmware 2, but I have no clue as to how to activate it. Everything looks the same as before, except the firmware says V2.

 

[Rodger1943]

Hi Alwyn, Don't worry about my request on how to find it, I have found it in the menu.

I must say that my menu is acting up a bit since I updated. I don't know whether its anything to do with the update, but it coincided with the update, so I suspect its behind it all. When I say acting up, you select the menu button, scroll through to a setting and before you can select it, the menu disappears and the camera reverts back to the shooting mode. With some difficulty, I managed to get to the bit where the animal eye setting was located and the camera jumped out of there straight back to the shooting mode. I'll see if this settles down, but if it persists, I won't be happy and you can't go back to Firmware V1.0.

I've just had another look at the menu and everything looks ok now. Fingers crossed.

--
Panasonic FZ1000 FZ200 Sony RX10 iv

 

[hwvlover]

Note that when I used Flexible Spot, once the Eye Focus locked on and started tracking, it immediately lost tracking when the eye went outside the Flexible Spot. Something to bear in mind.

Interesting, I have yet to try Eye AF on a moving dog but will as soon as it stops raining.
Based on my experience, with the standard lock-on modes, the subject is only tracked within the designated focus area, unless you are in wide you are invariably not using all the AF points.
Regards,
David
***************************************
Growing old is inevitable; growing up, however, is optional.
And I have opted out.

 

[Horacecoker]

Thanks for doing this testing Alwyn and posting the examples. You had a very good day, light wise for it!

In the past, mainly out of boredom, I've dabbled with human face and eye recognition from photos on my computer monitor and it worked very well. with that in mind, I decided to 'test' animal eye recognition using the images you posted out of curiosity.

I used my normal expand flexible spot I use for static birds at first and had mixed results. You are right that as soon as the eye is outside the box it ceases to track. Which to be fair is what I expected.

In the end, I simplified the test as follows.

I wanted the birds in the frame as a whole bird with space round them where possible so didn't have them at 100% view. (Except for the birds that were further away in your 2nd post). I set my RX10iv to 100mm equivalent focal length and positioned myself on my computer chair at a distance where the whole monitor just about filled the frame. I then focussed manually on the monitor, then switched to AF-C and wide focus area with animal eye AF switched on of course.

Now this is a thing that amazed me, without even touching the shutter button, if the camera recognised a bird's eye it put a white box on the eye and if I moved the camera it maintained the white box on the eye so obviously was tracking it. It did this across nearly the whole frame because I was in focus area, wide. If I half-pressed the shutter at any time during tracking the white box turned green.

If the camera did not recognise the eye in the first place, nothing I did by half-pressing the shutter to focus, made a blind bit of difference.

Out of the nine pictures you posted in your original post it recognised the eye on five of them; #2,#3, #6, #8 & #9. In your second post when the birds were further away it only recognised the eye in the last one but I'd zoomed to 100% view. Of the four images in your 3rd post, it nailed #1 & #4.

Of course, my 'test' is pretty worthless with these truly static bird images. It's out in the real world where it matters. Will I be using animal eye AF for birds? I very much doubt it unless they are really close. Nevertheless, this is a good start from Sony and hopefully, it can only get better!

Thanks again Alwyn.

 

[AlwynS]

Thanks for doing this testing Alwyn and posting the examples. You had a very good day, light wise for it!

Thanks David: always a treat to be out at Reifel, especially on a cool, crisp and at least partly sunny Fall day!

In the past, mainly out of boredom, I've dabbled with human face and eye recognition from photos on my computer monitor and it worked very well. with that in mind, I decided to 'test' animal eye recognition using the images you posted out of curiosity.

I used my normal expand flexible spot I use for static birds at first and had mixed results. You are right that as soon as the eye is outside the box it ceases to track. Which to be fair is what I expected.

Indeed: I had a bit of a "Duh!" moment myself when I first noticed it. Makes complete sense.

In the end, I simplified the test as follows.

I wanted the birds in the frame as a whole bird with space round them where possible so didn't have them at 100% view. (Except for the birds that were further away in your 2nd post). I set my RX10iv to 100mm equivalent focal length and positioned myself on my computer chair at a distance where the whole monitor just about filled the frame. I then focussed manually on the monitor, then switched to AF-C and wide focus area with animal eye AF switched on of course.

Now this is a thing that amazed me, without even touching the shutter button,

Interesting! I do not recall specifically noticing this. Do you have Pre-AF turned on by any chance? I have it turned off.

if the camera recognised a bird's eye it put a white box on the eye and if I moved the camera it maintained the white box on the eye so obviously was tracking it. It did this across nearly the whole frame because I was in focus area, wide. If I half-pressed the shutter at any time during tracking the white box turned green.

If the camera did not recognise the eye in the first place, nothing I did by half-pressing the shutter to focus, made a blind bit of difference.

Out of the nine pictures you posted in your original post it recognised the eye on five of them; #2,#3, #6, #8 & #9. In your second post when the birds were further away it only recognised the eye in the last one but I'd zoomed to 100% view. Of the four images in your 3rd post, it nailed #1 & #4.

Of course, my 'test' is pretty worthless with these truly static bird images. It's out in the real world where it matters. Will I be using animal eye AF for birds? I very much doubt it unless they are really close.

Agreed. The one advantage that I found was for close up birds with Wide focus area, I could lock the focus on the eye and then move the frame around to get my framing AND spot metering point where I want it to a) get the framing right and b) get the exposure as I want it. This is likely the only time I would use it.

Nevertheless, this is a good start from Sony and hopefully, it can only get better!

Agreed.

Thanks again Alwyn.

A pleasure: it was an interesting exercise.

 

[Kerusker]

Thanks for doing this testing Alwyn and posting the examples.

ditto

Now this is a thing that amazed me, without even touching the shutter button, if the camera recognised a bird's eye it put a white box on the eye and if I moved the camera it maintained the white box on the eye so obviously was tracking it. It did this across nearly the whole frame because I was in focus area, wide. If I half-pressed the shutter at any time during tracking the white box turned green.

I wondered about the white box(es) on the eye(s) some days ago too, but I couldn't find the settings today to repeat the effect 'white box on the eye'.

This was on photos on the wall with human eyes.

 

[Horacecoker]

In the end, I simplified the test as follows.

I wanted the birds in the frame as a whole bird with space round them where possible so didn't have them at 100% view. (Except for the birds that were further away in your 2nd post). I set my RX10iv to 100mm equivalent focal length and positioned myself on my computer chair at a distance where the whole monitor just about filled the frame. I then focussed manually on the monitor, then switched to AF-C and wide focus area with animal eye AF switched on of course.

Now this is a thing that amazed me, without even touching the shutter button,

if the camera recognised a bird's eye it put a white box on the eye and if I moved the camera it maintained the white box on the eye so obviously was tracking it. It did this across nearly the whole frame because I was in focus area, wide. If I half-pressed the shutter at any time during tracking the white box turned green.

If the camera did not recognise the eye in the first place, nothing I did by half-pressing the shutter to focus, made a blind bit of difference.

Interesting! I do not recall specifically noticing this. Do you have Pre-AF turned on by any chance? I have it turned off.

No, I never have pre-AF turned on. I think you are misunderstanding me. I didn't say the camera focussed on the bird's eye without touching the shutter button. I'm saying when the camera recognised an eye in one of your images, it placed a white box round it without any involvement from me. If it didn't recognise an eye nothing happened! To actually focus, the shutter button must be half-pressed and the white box turns green when focus is achieved. It must be stressed that if the animal/bird is so far out of focus as to be unrecognisable as an animal or bird, eye recognition will not take place. (Slightly out of focus no problem). That's why I manually focussed on the monitor before conducting the tests. My test was more about the ability of the camera to recognise birds eyes because unless it does, focussing on the eyes won't happen. It's a two-stage thing. Without the first stage happening you can forget about the second stage. That's why I used wide-area focus instead of my usual Expand Flexible Spot for static birds.

Out of the nine pictures you posted in your original post it recognised the eye on five of them; #2,#3, #6, #8 & #9. In your second post when the birds were further away it only recognised the eye in the last one but I'd zoomed to 100% view. Of the four images in your 3rd post, it nailed #1 & #4.

Of course, my 'test' is pretty worthless with these truly static bird images. It's out in the real world where it matters. Will I be using animal eye AF for birds? I very much doubt it unless they are really close.

Agreed. The one advantage that I found was for close up birds with Wide focus area, I could lock the focus on the eye and then move the frame around to get my framing AND spot metering point where I want it to a) get the framing right and b) get the exposure as I want it. This is likely the only time I would use it.

It's interesting to note that I've just repeated the test with my new A6600 & 16-55mm F2.8 lens. I set the lens on its maximum zoom (82.5mm equivalent). The results were identical!! This shows that the animal eye AF is as good (or bad) on the RX10iv as it is on the newly released top of the range Sony APS-C camera.

 

[Horacecoker]

Thanks for doing this testing Alwyn and posting the examples.

ditto

Now this is a thing that amazed me, without even touching the shutter button, if the camera recognised a bird's eye it put a white box on the eye and if I moved the camera it maintained the white box on the eye so obviously was tracking it. It did this across nearly the whole frame because I was in focus area, wide. If I half-pressed the shutter at any time during tracking the white box turned green.

I wondered about the white box(es) on the eye(s) some days ago too, but I couldn't find the settings today to repeat the effect 'white box on the eye'.

This was on photos on the wall with human eyes.

I've done some more tests and found that Human eye AF works differently. It is programmed to recognise faces first and foremost. So without any intervention from the photographer the camera puts a white box round a face, not an eye. It's only when you half-press the shutter that the box jumps to the eye but focusing is near-instantaneous so the white box will have turned green by that time it's on the eye! Move the camera with the shutter half-pressed and the green box remains on the eye but as soon as you take your finger off the shutter the tracking carries on but goes back to a white box on the face.

 

[hwvlover]

Thanks for doing this testing Alwyn and posting the examples.

ditto

Now this is a thing that amazed me, without even touching the shutter button, if the camera recognised a bird's eye it put a white box on the eye and if I moved the camera it maintained the white box on the eye so obviously was tracking it. It did this across nearly the whole frame because I was in focus area, wide. If I half-pressed the shutter at any time during tracking the white box turned green.

I wondered about the white box(es) on the eye(s) some days ago too, but I couldn't find the settings today to repeat the effect 'white box on the eye'.

This was on photos on the wall with human eyes.

I've done some more tests and found that Human eye AF works differently. It is programmed to recognise faces first and foremost. So without any intervention from the photographer the camera puts a white box round a face, not an eye. It's only when you half-press the shutter that the box jumps to the eye but focusing is near-instantaneous so the white box will have turned green by that time it's on the eye!

Animal eye AF does similar IF you have Animal Eye Display (4th item in the Face/Eye AF Set menu) turned on. No intervention from me and the white box is heading for Eszti's eye...she being one of my dogs. It seems that there is pre-focus going on here in this mode despite that being switched off.

With the Animal Eye Display function turned off there is no white box and pressing the shutter button brings up the green box instantly on the eye. So ISTBC but I would guess that pre-focus is once again occurring but having the display off means I am not seeing it.

Regards,
David
***************************************
Growing old is inevitable; growing up, however, is optional.
And I have opted out.

 

[Digital Shutterbug]

Animal eye AF does similar IF you have Animal Eye Display (4th item in the Face/Eye AF Set menu) turned on. No intervention from me and the white box is heading for Eszti's eye...she being one of my dogs. It seems that there is pre-focus going on here in this mode despite that being switched off.

I get a similar response, but it doesn't actually focus until I press the BBF button. The square is able to lock onto a face that is not much more than a fuzzy blob, barely identifiable as a face. The camera will initiate focus if I push the eye AF button, the button on the side of the lens in my configuration. This happens without depressing the BBF. Of course, if it doesn't;t detect an eye, pushing the eye AF button does not start focus, and pushing the BBF is required. I don't mind this action. I do not want a true pre-focus as I detest that mode.

With the Animal Eye Display function turned off there is no white box and pressing the shutter button brings up the green box instantly on the eye. So ISTBC but I would guess that pre-focus is once again occurring but having the display off means I am not seeing it.

Regards,
David
***************************************
Growing old is inevitable; growing up, however, is optional.
And I have opted out.

 

[Digital Shutterbug]

No, I never have pre-AF turned on. I think you are misunderstanding me. I didn't say the camera focussed on the bird's eye without touching the shutter button. I'm saying when the camera recognised an eye in one of your images, it placed a white box round it without any involvement from me. If it didn't recognise an eye nothing happened! To actually focus, the shutter button must be half-pressed and the white box turns green when focus is achieved. It must be stressed that if the animal/bird is so far out of focus as to be unrecognisable as an animal or bird, eye recognition will not take place. (Slightly out of focus no problem). That's why I manually focussed on the monitor before conducting the tests. My test was more about the ability of the camera to recognise birds eyes because unless it does, focussing on the eyes won't happen. It's a two-stage thing. Without the first stage happening you can forget about the second stage. That's why I used wide-area focus instead of my usual Expand Flexible Spot for static birds.

This is the exact way my camera is working. No pre-focus. It simply recognizes the subject and marks it, waiting on me to initiate focus. I'm surprised how out of focus a face can be and the camera still recognizes it as a face. There are limits though.

It's interesting to note that I've just repeated the test with my new A6600 & 16-55mm F2.8 lens. I set the lens on its maximum zoom (82.5mm equivalent). The results were identical!! This shows that the animal eye AF is as good (or bad) on the RX10iv as it is on the newly released top of the range Sony APS-C camera.

 

[hwvlover]

Animal eye AF does similar IF you have Animal Eye Display (4th item in the Face/Eye AF Set menu) turned on. No intervention from me and the white box is heading for Eszti's eye...she being one of my dogs. It seems that there is pre-focus going on here in this mode despite that being switched off.

I get a similar response, but it doesn't actually focus until I press the BBF button. The square is able to lock onto a face that is not much more than a fuzzy blob, barely identifiable as a face. The camera will initiate focus if I push the eye AF button, the button on the side of the lens in my configuration. This happens without depressing the BBF. Of course, if it doesn't;t detect an eye, pushing the eye AF button does not start focus, and pushing the BBF is required. I don't mind this action. I do not want a true pre-focus as I detest that mode.

With the Animal Eye Display function turned off there is no white box and pressing the shutter button brings up the green box instantly on the eye. So ISTBC but I would guess that pre-focus is once again occurring but having the display off means I am not seeing it.

Regards,
David
***************************************
Growing old is inevitable; growing up, however, is optional.
And I have opted out.

Thank you Steve, you are right. Unless pre-focus (I have never had that on) means before focus my description is of course wrong. What is happening is something of a "likely target" acquisition, focus then comes when I press the back button.
Regards,
David
***************************************
Growing old is inevitable; growing up, however, is optional.
And I have opted out.

 

[tbcass]

Sony makes no claim that the new EYE AF tracking algorithms works on birds with any of their cameras. Why did you expect it would? It seems to work on dogs.

--
Tom

 

[tbcass]

Note that when I used Flexible Spot, once the Eye Focus locked on and started tracking, it immediately lost tracking when the eye went outside the Flexible Spot. Something to bear in mind.

Interesting, I have yet to try Eye AF on a moving dog but will as soon as it stops raining.
Based on my experience, with the standard lock-on modes, the subject is only tracked within the designated focus area, unless you are in wide you are invariably not using all the AF points.

Did you try lock on expand flexible spot with AF tracking? When I use that mode the camera will track all over the screen.

 

[AlwynS]

Sony makes no claim that the new EYE AF tracking algorithms works on birds with any of their cameras. Why did you expect it would? It seems to work on dogs.

Actually I had ZERO expectations. I started this thread with the comment "The efficacy or not of the new firmware 2.0 release for the RX10 IV has been discussed in some other threads." This was largely in response to this thread that had a lot of speculation.

As regards my own expectations: perhaps I should have quoted myself from THAT thread at the start of THIS thread for perspective. For ease of reference: a few of my pertinent quotes from that thread:

Quote 1:

"As regards Sony's Animal Eye AF: I have done a quick Google search/scan on the topic. And I frequently came across statements that indicate that Sony has not satisfactorily implemented eye AF for birds, even by their own admission. On Sony's own support sites, comments like (from the RX10IV firmware site) "To detect animal eyes, arrange the composition so that both eyes and the nose of the animal are within the angle of view. Once you focus on the animal’s face, the animal’s eyes will be detected more easily." and "When the subjects have cat- or dog-like faces and both eyes are clearly visible, the camera can easily detect the eye" are not confidence-inspiring for BIF. They also specifically note that "Animals that do not have cat- or dog-like faces" are hard to detect."

This for me pretty much set my expectations. Dogs and cats yes. Birds no. Curiously enough you actually responded to that post of mine sharing my skepticism....

Quote 2:

"If you read the updated manual and other Sony blurb on the topic, it becomes pretty clear that their objective was for eye AF on slow moving "dog- and cat-like" animals that are not far away: pretty much the antithesis of fast moving, small, single eyed, "undoglike" BIF at a distance."

Ergo: NOT meant for birds, reinforcing my lack of expectations.

Quote 3: by Stephen Ingram, the OP of the noted thread. This in response to my post where I noted that his thread was regarding general tracking, NOT Eye AF for birds.

"This is correct. Animal eye tracking is specifically NOT implemented for birds on any Sony camera. I had a long discussion with one of the Sony people about what it would take and it seems to be beyond what the want to attempt right now. My comments were on tracking auto focus not animal eye tracking"

Quote 4: Also in that thread I posted some "quick and dirty" test shots which indicated that it seems to work on dogs (as expected) but not on birds (also as expected). I said:

"So a VERY small test run but what there is is not promising for bird eye AF (as Sony suggested)."

And yes, in retrospect this last sentence is ambiguous. I should have made it more clear that Sony made it quite clear that Animal Eye AF was aimed at dogs and cats etc, NOT birds. So it was no surprise to me when it did not work well on birds.

So to reiterate: going into this "test", I had zero expectations that the Eye AF would work on birds. To be honest, what it did achieve was actually better than I expected.

 

[Digital Shutterbug]

Interesting, I have yet to try Eye AF on a moving dog but will as soon as it stops raining.
Based on my experience, with the standard lock-on modes, the subject is only tracked within the designated focus area, unless you are in wide you are invariably not using all the AF points.

Did you try lock on expand flexible spot with AF tracking? When I use that mode the camera will track all over the screen.

Unfortunately, you cannot select Lock-On (anything) with Animal eye AF selected. You can with Human eye AF selected, but not Animal. If you have Lock-On selected, and then switch to Animal eye AF, the camera automatically reverts to Expand Flexible Spot (without telling you). Switching back to Human eye AF, the camera will go back to Lock-On (again without notification). You have to watch the focus symbol, or menu dive, to see what is happening.

The one notification you get is trying to switch to Lock-On while currently in Animal eye AF. Lock-On is greyed out, if you ignore it and attempt to select it anyway, it notifies you that it can't be selected while in Animal setting.

 

[tbcass]

Interesting, I have yet to try Eye AF on a moving dog but will as soon as it stops raining.
Based on my experience, with the standard lock-on modes, the subject is only tracked within the designated focus area, unless you are in wide you are invariably not using all the AF points.

Did you try lock on expand flexible spot with AF tracking? When I use that mode the camera will track all over the screen.

Unfortunately, you cannot select Lock-On (anything) with Animal eye AF selected.

I did not know that but hwvlover said he has not tried it on moving dogs. He also said the subject is only tracked within the designated focus area with the standard lock-on modes which led me to believe he wasn't talking about animal eye AF.

You can with Human eye AF selected, but not Animal. If you have Lock-On selected, and then switch to Animal eye AF, the camera automatically reverts to Expand Flexible Spot (without telling you). Switching back to Human eye AF, the camera will go back to Lock-On (again without notification). You have to watch the focus symbol, or menu dive, to see what is happening.

The one notification you get is trying to switch to Lock-On while currently in Animal eye AF. Lock-On is greyed out, if you ignore it and attempt to select it anyway, it notifies you that it can't be selected while in Animal setting.

That is interesting and informative.