AF for birds in flight with the Sony A9 and the 200-600 zoom

In prior posts I have presented data about AF for birds in flight with the Sony A9 and the 200-600 zoom. I have shown that in challenging conditions , the system did not perform at the 90% accuracy level some claim. In this post I want to look at some aspects of the system to explore why this might be so.

How the PDAF system affects AF for birds in flight with the Sony A9

All pro DSLRs and most high end mirrorless cameras use Phase-Detect Auto Focus (PDAF) sensors. PDAF sensors are basically tiny, simple rangefinders — light from each side of the lens is split into two paths, and the two paths are projected onto small linear sensors. The difference between the pattern of light and dark is analysed, and the amount of front or back focus instantly calculated. That’s why phase-detect autofocus is so much faster than contrast-detection as implemented in most point & shoot cameras — with that, the lens has to seek back and forth to find the point of most contrast. It’s generally acknowledged that PDAF is essential for birds in flight (BIF) to be able to quickly focus on the lens without hunting.

The A9 autofocus system includes 693 phase-detection points which cover 93% of the image area. In addition to these 693 PDAF points, there are also 425 contrast-detection autofocus points (CDAF). It appears the CDAF points are used for Single AF and areas such as Flexible Spot, Centre or Expand Flexible Spot, as well as for low light. Continuous AF apparently uses PDAF until f9, when it switches to CDAF. A combination of low light, and an f9 aperture, as in the 200-600 plus 1.4TC (which results in an f9 aperture), probably guarantees CDAF will be used, which may account for the very poor focussing performance of this combination in low light.

PDAF sensor orientation

All pro DSLRs (and one mirrorless camera system) use cross-type PDAF sensors. A cross-type sensor is simply two linear sensors crossing each other (making a cross shape), which is superior because it can work with both vertical and horizontal light patterns.

However, all mirrorless camera systems (except one), use linear PDAF sensors and not cross-type, which means that they are vastly better at picking up focus in one axis compared to the other. In the case of the Sony A9, the sensor is arranged vertically, which means that it is good at acquiring focus on vertical contrast, but poor at purely horizontal contrast.

Here is an excellent video from one of the best technical gurus out there, Gary Friedman, showing exactly what happens. Pick it up at around 0:55s.

By the way, Gary has been a Sony proponent and expert for 10 years or more, and I was first set on a path of serious photography by attending one of his courses in 2012 with my brand new Sony cameras (at that time I had the Nex 6 and the RX-100).

Does this affect the BIF focussing of the A9? Well there are many situations where the main items of contrast in a BIF shot are horizontal. Look at this series of photos of a slow moving vulture, moving from right to left around 50m away. After the first 3 shots (when I told the camera where to look), the A9 completely fails to lock focus, and is hunting all over the frame, until the bird passes over a chair with some vertical contrast. As soon as it sees the chair, the A9 gets focus lock, and although it is not brilliant, performs far better thereafter. This was an incredibly bad performance from the A9, and I think it it plausible to attribute this in part to its simple, one-axis PDAF system. The other explanation is that the camera switched to CDAF (see above), which is not very effective for BIF.

The A9 and the Olympus Pro cameras

I mentioned above, that there was one mirrorless camera system that used cross-type PDAF points. Which system might that be? It turns out the Olympus Pro cameras, the EM1 ii, EM1iii and EM1x are the only mirrorless cameras with all cross-type sensors.

Paraphrasing an Olympus engineer “The AF sensor array layout on the image sensor – which is unique to Olympus – is designed to detect any vertical, horizontal, or diagonal line patterns on subjects, and to find a defocused subject…. The PDAF system can detect vertical and horizontal line patterns equally, allowing the camera to detect and focus on subjects in the foreground….

I think it is entirely plausible that the reason that the EM1x in my tests was slightly better than the A9 for difficult BIF situations is because of the much more sophisticated PDAF array it uses.

Sony A9 and the 200-600 zoom – focus acquisition

The AF for birds in flight with the Sony A9 and the 200-600 zoom was worst in situations where you have to immediately focus and shoot a bird whose position is changing wildly and quickly and you have maybe 1-3 seconds to get the picture. In this situation, which is common when a single bird is being shot in the wild, the A9 was slow to gain focus and on many occasions did not get the shot (39% accuracy cf 45% on the EM1x). 

This can probably be put down in part to the 200-600 lens. This is a fantastic piece of glass, and superb value. However, it is not a GMaster lens, and apparently has only one focus motor, as opposed to two used by the other GMaster telephoto lenses such as the 400mm f2.8 and the 600mm f4. It is slower to acquire focus than it’s more expensive compatriots.

The Olympus 300mm prime however is a different kettle of fish. The lens elements are smaller and lighter and and faster to move, and it is constructed as a pro lens. There are no comparative tests of focus acquisition for telephoto lenses that I have been able to find, but this review rates the Olympus 300mm f4 as fast for focus acquisition as the Canon 500mm f4, regarded by many as the fastest and best birding lens in the world. I think it is this, plus the acknowledged speed of the EM1x focus acquisition that led to the advantage that I saw in very fast reaction shots.

Sony A9 and the 200-600 zoom – Inertia

The 200-600 is by no means the heaviest telephoto in the Sony range, but at 32cm long, and 2.1 kg in weight, it is substantially larger and heavier than the Olympus 300mm f4 which is 1.3kg and 22 cm long. Size and inertia are important when you have to flick the camera around to capture fast action.

At the HCT afternoon sessions, the climax of the kite flying display is a feeding frenzy where chicken pieces are catapulted vertically up in the air, past flying kites, and they fly upwards to chase them, and jostle each other out of the way in the process. They take maybe 1 second to find the chicken piece, and can remain static in the air for a millisecond, before swooping down. Physically swinging the Sony A9 and the 200-600 zoom around is a fraction slower than the Olympus, and combined with a slightly slower lens, that’s all it takes to miss the shot.

Conclusion

There is good data to suggest that the A9 and the 200-600 does not achieve 90% BIF focus accuracy in difficult shooting conditions There are three plausible reasons, listed above, why this may be so, and why the EM1x camera can perform slightly better than the A9 in the same situations.