As much as I love Sony and A1 but the AI behind this feature is not 100% reliable. It will be improving as time going forward. For static, it is 100% on the dot. When it is in flight, there are so many variables to add in, combine with 30 or 20 fps, you can't expect it to be perfect.
If you compare this features with A9/A9ii, it is a massive upgrade for sure.
You want 100% hit rate, try subject tracking flex S. It works ONLY if your tracking/panning can keep up with it.
"To truly test the bird eye AF of the newly launched Sony a1, I needed to find a moving bird against a distracting background that would normally confuse the camera's AF system into focusing on the background. Sounds easier than it was, as I first had to overcome my natural reflex of finding birds and animals against nice clean out-of-focus backgrounds. This image is perfect to demonstrate the ridiculous accuracy of the bird eye autofocus of the Sony a1: Not only is the background close and cluttered (photographing the subject further from the background would have made the background more out of focus and made the whole test easier for the camera), but the subject is relatively low contrast against it. The random scattering of snow covered branches is harder for most autofocus systems to negate when compared to the predictable vertical tree trunks on the edge of a deciduous forest. It was remarkable to see the camera lock onto the eye and follow it through a sequence of 40 images that were all razor-sharp."
arbitrage wrote:
One last thing to note...it is valuable to keep a custom key (I currently use the left arrow) to Toggle (toggle only available on A9II via FW and now A1 (maybe A7SIII)) the Face/Eye Detection On/Off. Maybe the new system is much improved but the A9II system often gets confused with patterns and once it thinks something is an eye (which in this case it isn't) it will only focus on that. You sometimes need a quick toggle to turn off Face/Eye detect temporarily. Anyways, hopefully the A1 is much improved and false positives will be minimized or even eliminated.
I have Face/eye detect set to C1 , for perched birds or Ducks I'll quickly toggle it on to see if it grabs the eye, if not I just tap C1 to turn off and use normal Flex spot or centre .
The video Alex uploaded certainly looks like the A1 eye detect is a lot better than the A9ii , interesting to see the variation of birds it works with
Alex Phan wrote:
As much as I love Sony and A1 but the AI behind this feature is not 100% reliable. It will be improving as time going forward. For static, it is 100% on the dot. When it is in flight, there are so many variables to add in, combine with 30 or 20 fps, you can't expect it to be perfect.
If you compare this features with A9/A9ii, it is a massive upgrade for sure.
You want 100% hit rate, try subject tracking flex S. It works ONLY if your tracking/panning can keep up with it.
Even if its not perfect for BIF but 100% for static sounds good to me
"To truly test the bird eye AF of the newly launched Sony a1, I needed to find a moving bird against a distracting background that would normally confuse the camera's AF system into focusing on the background. Sounds easier than it was, as I first had to overcome my natural reflex of finding birds and animals against nice clean out-of-focus backgrounds. This image is perfect to demonstrate the ridiculous accuracy of the bird eye autofocus of the Sony a1: Not only is the background close and cluttered (photographing the subject further from the background would have made the background more out of focus and made the whole test easier for the camera), but the subject is relatively low contrast against it. The random scattering of snow covered branches is harder for most autofocus systems to negate when compared to the predictable vertical tree trunks on the edge of a deciduous forest. It was remarkable to see the camera lock onto the eye and follow it through a sequence of 40 images that were all razor-sharp."
Hold on here just a minute.
I'm a sony fanboy, currently own 6 sony bodies including a9/a7riv.
I'm sure the alpha1 will be awesome. Alex/Galer have already shown so.
But I bet even Dodds would admit a Great Grey is a large and slow moving bierd.
It doesn't take the most fantastic BEAF (thanks Geoff for that acronym) to get a good shot of a GG.
Here's one with the 7Dii and most of us know the 7Dii was not the greatest.
Personally I want to see what is done with the Peregrines, and the Swallows. Beauport Outtake 1 by MedicineMan4040, on Flickr
MedicineMan404 wrote:
Hold on here just a minute.
I'm a sony fanboy, currently own 6 sony bodies including a9/a7riv.
I'm sure the alpha1 will be awesome. Alex/Galer have already shown so.
But I bet even Dodds would admit a Great Grey is a large and slow moving bierd.
It doesn't take the most fantastic BEAF (thanks Geoff for that acronym) to get a good shot of a GG.
Here's one with the 7Dii and most of us know the 7Dii was not the greatest.
Personally I want to see what is done with the Peregrines, and the Swallows. https://live.staticflickr.com/7305/16344980809_941f386b76_h.jpgBeauport Outtake 1 by MedicineMan4040, on Flickr
Your background doesn't look as busy, imo. It would interest me, whether you also got 40 images in a row in perfect focus on the eye at 20fps and with similar DOF?
Of course you can always show images which demonstrate, that a similar image is possible, but we never know whether it was luck, a cherry picked image, or whether you get those shots all the time without fail with busy backgrounds. That doesn't make your shot worse (a nice image btw).
Now after having a good 4 months of Canon R5 BEAF with BIF experience I will say that BEAF is likely going to work more often on larger and slower BIF....that won't come as a surprise to anyone. But I will say that BEAF is most useful on larger BIF because it is those where you start to run into near wing focus with traditional AF Area modes like Zone or Wide. BEAF will likely not work on a fast Bufflehead and certainly isn't going to work on a swallow. But for those types of birds a regular Sony Zone or Wide on an A9II already does really well...subject is smaller in frame and DOF covers most or all of it in most shooting situations. I don't need BEAF for the ducks and swallows...what I need is A9II AF at 50MPs. And the swallows won't hurt to have 30FPS. I'll probably actually turn BEAF off shooting swallows just to not overload the camera
BEAF on the R5 has been most valuable on non-BIF situations. It can help react just a little bit faster to a fleeting perching moment. But most valuable to me has been having it help me in my awkward shooting position I use for water's edge, flip-screen photography with the big 600GM lens (I've borrowed a few Canon big lenses for the R5). My technique is to rest the camera on the toe of my left shoe, hold the top edge of my lens hood with my outstretched left hand and operate just the shutter-button AF with my right hand from above (not properly gripping the camera). That all leads to a little bit of an unstable situation. Canon BEAF has made it so much easier to not have to worry much about moving my Sony Zone focus area around and to just concentrate getting some sort of good composition.
What I'm most curious about when I get to compare A1 to R5 BEAF is if the A1 can pick out the bird when it is in odd poses. Yesterday I was shooting nuthatches hanging down vertically and then tucking themselves up with head obscured and the R5 was still recognizing them as birds even when it couldn't see the eye and when the bird was not in any traditional bird shape. I don't think it will matter much if the A1 can do that or not because the A1 will just drop back to Zone or RTT where as the R5 drops back to body detect and draws a larger box over the entire bird. End result will be similar. The A1 doesn't seem to have the middle ground like the R5 does....R5 goes from dancing dots to rectangle on entire bird to square on the head to small square on the eye. The size and shape of the rectangle and squares vary as it gets more confidence in what it is detecting. The A1 seems to just go from dancing dots (or RTT icon) to eye square (with slight size variation) but nothing in between.
arbitrage wrote:
Canon BEAF has made it so much easier to not have to worry much about moving my Sony Zone focus area around and to just concentrate getting some sort of good composition.
I use the Sony RTT in that scenario. However, you still have to find the critter to initiate it.
arbitrage wrote:
Now after having a good 4 months of Canon R5 BEAF with BIF experience I will say that BEAF is likely going to work more often on larger and slower BIF....that won't come as a surprise to anyone. But I will say that BEAF is most useful on larger BIF because it is those where you start to run into near wing focus with traditional AF Area modes like Zone or Wide. BEAF will likely not work on a fast Bufflehead and certainly isn't going to work on a swallow. But for those types of birds a regular Sony Zone or Wide on an A9II already does really well...subject is smaller in frame and DOF covers most or all of it in most shooting situations. I don't need BEAF for the ducks and swallows...what I need is A9II AF at 50MPs. And the swallows won't hurt to have 30FPS. I'll probably actually turn BEAF off shooting swallows just to not overload the camera
BEAF on the R5 has been most valuable on non-BIF situations. It can help react just a little bit faster to a fleeting perching moment. But most valuable to me has been having it help me in my awkward shooting position I use for water's edge, flip-screen photography with the big 600GM lens (I've borrowed a few Canon big lenses for the R5). My technique is to rest the camera on the toe of my left shoe, hold the top edge of my lens hood with my outstretched left hand and operate just the shutter-button AF with my right hand from above (not properly gripping the camera). That all leads to a little bit of an unstable situation. Canon BEAF has made it so much easier to not have to worry much about moving my Sony Zone focus area around and to just concentrate getting some sort of good composition.
What I'm most curious about when I get to compare A1 to R5 BEAF is if the A1 can pick out the bird when it is in odd poses. Yesterday I was shooting nuthatches hanging down vertically and then tucking themselves up with head obscured and the R5 was still recognizing them as birds even when it couldn't see the eye and when the bird was not in any traditional bird shape. I don't think it will matter much if the A1 can do that or not because the A1 will just drop back to Zone or RTT where as the R5 drops back to body detect and draws a larger box over the entire bird. End result will be similar. The A1 doesn't seem to have the middle ground like the R5 does....R5 goes from dancing dots to rectangle on entire bird to square on the head to small square on the eye. The size and shape of the rectangle and squares vary as it gets more confidence in what it is detecting. The A1 seems to just go from dancing dots (or RTT icon) to eye square (with slight size variation) but nothing in between....Show more →
Geoff, that an excellent point you made about using the BE-AF for perched or slow moving targets where you want the camera to find the eye fast and with minimal effort on us. I can see myself using the BE-AF in the A1 the way you described it. I will try the BE-AF in that camera for BIF. If it works well, great; otherwise no big deal, I have taken plenty of BIF using the A9 II or even the A7r IV to see that the Zone-AF works really well for those conditions. By all accounts, the A1 should work even better than those 2 cameras, especially better than the A7r IV. So, I am perfectly fine with that.
We will be able to evaluate this camera in 2 weeks first hand and I am looking forward to that!
AGeoJO wrote:
Geoff, that an excellent point you made about using the BE-AF for perched or slow moving targets where you want the camera to find the eye fast and with minimal effort on us. I can see myself using the BE-AF in the A1 the way you described it. I will try the BE-AF in that camera for BIF. If it works well, great; otherwise no big deal, I have taken plenty of BIF using the A9 II or even the A7r IV to see that the Zone-AF works really well for those conditions. By all accounts, the A1 should work even better than those 2 cameras, especially better than the A7r IV. So, I am perfectly fine with that.
We will be able to evaluate this camera in 2 weeks first hand and I am looking forward to that! ...Show more →
Exactly how I feel as well... I really hope I have the camera for warbler season because those guys just love to mess with me by moving at the last second beforeI can get the focus point exactly where I want.
I have just noticed that a7info reports FocusDIstance as a parameter under the Lens section of Exif data. I can't find it in the Exif using Exiftool. Does anyone know where it comes from?
EDIT: I just answered my own question! Using ExifToolGUI, there is an option to turn on "Show Composite Tages in view ALL". With that switched on, you can see a tag FocusDistance2 under ALL. It doesn't entirely match the value show in a7info but then this is a a derived or calculated value. I'm not sure which is most accurate. In Alex' Heron example, the values shown in a7info and ExifToolGUI diverge as the bird gets closer.
AGain with the heron example, the FocusDIstance(2) attribute either increments down between frames as the bird gets closer or sometimes keeps the same value between frames. I'm not sure under what circumstances that happens but it is at least good to see a decreasing distance trend between images. My Olympus cameras were frustrating in that the FocusDistance parameter wavered up and down around the true value and this was clearly visible in the images.
The Osprey take-off sequence is less satisfying in that the value rarely changes. Perhaps this indicates that FocusDIstance2 is a very roughly derived value. ANyone know what it is derived from?
somersettr wrote:
I have just noticed that a7info reports FocusDIstance as a parameter under the Lens section of Exif data. I can't find it in the Exif using Exiftool. Does anyone know where it comes from?
It's called FocusDistance2 in ExifTool but it reports a bit different value from what I've seen...
"To truly test the bird eye AF of the newly launched Sony a1, I needed to find a moving bird against a distracting background that would normally confuse the camera's AF system into focusing on the background. Sounds easier than it was, as I first had to overcome my natural reflex of finding birds and animals against nice clean out-of-focus backgrounds. This image is perfect to demonstrate the ridiculous accuracy of the bird eye autofocus of the Sony a1: Not only is the background close and cluttered (photographing the subject further from the background would have made the background more out of focus and made the whole test easier for the camera), but the subject is relatively low contrast against it. The random scattering of snow covered branches is harder for most autofocus systems to negate when compared to the predictable vertical tree trunks on the edge of a deciduous forest. It was remarkable to see the camera lock onto the eye and follow it through a sequence of 40 images that were all razor-sharp."
While the Bird Eye Autofocus is quite impressive and better than anything else that I have tried, I did note in my initial review that it did not work for me on small dark birds: the focus locked-onto the shoulder. I expect this will improve with firmware updates that Sony is quite proactive in providing.
There is a lot more contrast in your image, with a much different background.
If you care to take the time, have a read through the a1 posts on my blog (and the comments): The bird eye AF is not perfect, and I am careful to say that and mention that I have limited time/opportunity and limited species to photograph. I mentioned a problem I had when the autofocus locked onto a dark bird's shoulder...
Chris Dodds wrote:
While the Bird Eye Autofocus is quite impressive and better than anything else that I have tried, I did note in my initial review that it did not work for me on small dark birds: the focus locked-onto the shoulder. I expect this will improve with firmware updates that Sony is quite proactive in providing.
Yeah, since the system basically have to analyze an image of the current subject in a split second (it will thus always be a bit behind the present moment), it is not surprising that finding a dark eye on a dark head can be tricky. What would be nice is a fallback where the system will recognize the head of the bird at least, a kind of silhouette recognition system.
Stoffer wrote:
Yeah, since the system basically have to analyze an image of the current subject in a split second (it will thus always be a bit behind the present moment), it is not surprising that finding a dark eye on a dark head can be tricky. What would be nice is a fallback where the system will recognize the head of the bird at least, a kind of silhouette recognition system.
That is what the R5 does, falls back to head and then falls back to the silhouette/body. A1 seems like it will just fall back to the RTT if you are using the Tracking modes or whatever your base AF mode is.
Chris Dodds wrote:
While the Bird Eye Autofocus is quite impressive and better than anything else that I have tried, I did note in my initial review that it did not work for me on small dark birds: the focus locked-onto the shoulder. I expect this will improve with firmware updates that Sony is quite proactive in providing.
That is fine. I personally will never use that feature, since I photograph weddings, but I like the possibility it offers for people shooting BIF. For me there are many other great features in the A1 making this camera a must have.
