Sony A9 II for BIF, initial impressions and comparison vs. Nikon/Canon

Sony A9 II for BIF, initial impressions and comparison vs. Nikon/Canon

For the past couple of months I have been shooting with the Sony A9 II mirror-less interchangeable lens camera (MILC). I have been receiving many requests to compare the Sony with the pro DSRL’s given my prior experience with both Nikon D5 and Canon 1D series camera and 600mm lenses. While I have not been able to do extensive photography due to the COVID-19 pandemic I have shot enough with the Sony to form an opinion of the camera compared to my current Nikon and previous Canon setups.

I will cover the differences in ergonomics, menu system, image quality etc. in a detailed review of the A9 II later. Here I am only going to talk about the fundamental difference in focus tracking for bird in flight subjects.

Lens first 

It is meaningless to talk about birds in flight and serious bird photography in general without looking at super-telephoto lenses first, in particular the 600mm f/4 lenses, aka “the Big Six”, a bird photographer’s primary weapon of choice. Without a super-telephoto lens, a camera system falls short for serious BIF work in my view. The Sony 600mm f/4 GM (and the Canon 600mm f/4 MK III) are both about 2lbs lighter than the Nikon 600mm f/4E FL. While I can hand hold the Nikon all day long and hike with it too, a lighter lens is always better for more accurate tracking and placement of the bird. It causes less fatigue and it’s also easier to handle from a vehicle. The Sony lens is so light I can grab it by the hood and due to the reduced lever action, I can precisely pan with the subject. Something that can’t be done with the heavier lens.

All three bare lenses are tack sharp on any camera and equally tack sharp with the 1.4X TC. There is no point in comparing them in this regard. However there are suddenly significant differences when it comes to the 2X TC. The Sony 2X TC is not just tack sharp but it is also very thin and light. It doesn’t add much weight to the rig or change the center of gravity of the rig. In comparison the Canon 2X III is very thick and somewhat heavier, it will extend the center of gravity further from the camera body. The Canon 2X III is reasonably sharp but it is not as tack sharp as the Sony 2X which is a newer design. The Canon 2X TC III can show a bit fringing (chromatic aberration) when used with Canon’s 600mm f/4, especially when shooting far subjects this combo is used for. The CA can shows in some subjects such as terns, white-tailed kites, skimmers with low contrast white feathers, but overall it is more than usable. The Sony 2X TC on the other hand is almost perfect. On top of that Canon’s AF at f/8 is a hit and miss and not something that can be used in all conditions for BIF. The Sony, on the other hand, appears to take no hit from the 2X TC in terms of AF (more on this later). In fact 90% of the images I made with the A9 II were with the 600GM + 2X TC (1200mm). The Sony AF makes it incredibly easy to shoot at 1200mm to the point that you don’t think about it. The 2X TC can be literally glued to the lens and forgotten.

The Nikon 2X TC is unfortunately outright soft, it shows strong fringing at pixel level and its aging optics is no match for the competition. I advise against using it, simply swap the D5 for the D500/D850 when more reach is needed and you will get day and night better results than the 2X on a D5.

because of these factors Sony wins in the optics category, in terms if weight and the maximum reach (1200mm) with full blown AF performance and sharp files. Canon 600 III gets the second place because of weight compared to Nikon.

the table below compares the lenses

 

 

A revolution in AF technology, not an evolution 

In order to understand the differences between MILC and DSLR cameras we need to explain how the DSLR camera focuses first”

DSLR stands for Digital Single Lens Reflex, basically an old SLR camera with a digital image sensor in place of the film roll plus an LCD monitor and loads of modern electronics.  In these cameras a mirror is used to reflect the light incident by the lens towards a prism which then reflects the light towards the viewfinder and the eye. So how does the camera auto focus? There are a few spots on the main mirror (near the center) where the mirror is made transparent by etching of the reflective coating. A fraction of the light goes through these spots. A secondary mirror behind the main mirror is placed at angle to reflect the light towards the base of the camera body. This secondary mirror is called the “sub-mirror“. The sub-mirror is attached to and moves with the main mirror. When the mirror is up for exposure, no light can get to the AF sensor.

The light then  passed through a group of small lenses before being reflected by a third fixed mirror at the base of the camera. This mirror bounces the light towards another group of small lenses which will spread it on the AF sensor. The function of the tiny lenses in the light path is to make the optical path between the lens and the AF sensor equal to the path between the lens and the image sensor. So when the image is in focus on the AF sensor, it will also be in sharp focus on the image sensor. This system has essentially been the same since the introduction of Auto Focus (AF) SLR cameras back in late 1970’s in the film days.  Although many subtle and evolutionary improvements have been made through the decades, the concept of the system remains unchanged. The system is very complex, has both fixed and moving parts and has to be aligned and calibrated perfectly to the tee. This is part of the manufacturer know how and partially responsible for 6000+ USD price tag of a pro DSLR body. It has been getting  the job done, most of the time.

The schematics below (source Canon CPS 1DX Mark II publication) shows the light path to the AF sensor in the Canon 1D X series camera. The 2nd image below shows a cutaway  of the the Nikon D4 with the entire AF system in this camera (borrowed from the Digital Camera World)

 

Optical path from the lens to the AF sensor shown for the Canon EOS-1D X MK II. Only a fraction of light is transmitted to the sub-mirror where it is reflected towards the base of the camera body. After bouncing from a 3rd mirror and the image forming lens, the light hits the AF sensor. Source: Canon publication

 

 

Cutaway of Nikon D4 showing the entire AF system.  Source: https://www.digitalcameraworld.com/

As you can imagine since most of the light has to go to the viewfinder, only a fraction will make it to the AF sensor, projecting a relatively dim circle on the AF sensor. This is why DSLR cameras can only focus with a max. aperture of f/8 or faster and have some limitations when focusing at f/8 or even f/5.6 relative to f/4 or f/2.8. With each stop loss of light the illuminated circle on the AF sensor get dimmer and smaller, although the scene may not appear so dim. It’s remarkably they can still focus at f/8. This is also the reason why the outer focus points become inactive or lose precision at f/8.

The other issue is the AF coverage, the fraction of light that makes it thought the AF system is not able to illuminate a circle large enough to cover the corners of a full-frame image. That’s why the AF sensor can only cover  the central part of the image (typically the center 25% area).  The image below shows the evolution of the AF sensors since the almost decade-old EOS-1DX and Nikon D4 to the current EOS-1DX III as well as the Nikon D6. As you can see the AF grid’s outer coverage hasn’t really changed much given this fundamental limit.

 

 

Evolution of DSLR AF sensor in the past decade. The number of points and density has increased, the coverage remains at approx. 25% area of the frame. (clustered at the center)

 

AF coverage for the EOS-1D X Mark III is roughly 24% of the full frame area, i.e. the shaded area divided by the total viewfinder area shown by the black frame.

 

The MILC cameras on the other hand do not require an AF sub-system. There is no dedicated AF sensor and no optical sub-system associated with it. The main image sensor is used for focusing as well. The advantage of this method is that the AF sensor has all the light projected by the lens available to it at all times. The Sony A9 II can focus all the way down to f/13 and does not show a noticeable hit in AF when attaching a TC to a prime lens. Also the entire area of the image sensor can be used for focus detection so the AF coverage is virtually 100% of the frame.

Initially the MLIC cameras could AF by contrast detection (CD) only. This process is accurate but very slow and causes AF hunting, it is not suitable for action shooting at a high frame rate. But today the sensors perform phase-detect (PD) AF by partially masking the micro-lens on the pixels that are used for the AF system to create the right and left split images. The phase detection is coupled with contrast detection to offer both speed and accuracy.

Another difference is the number of focus points available  to the camera for focus detection. As explained above the DSLR AF system has to make due with a much dimmer image therefore the AF sensor needs to have very large pixels to read the contrast. In the past the AF sensors were simple fabricated line sensors as seen in the EOS-1D X Mark II’s AF sensor image below. This meant very few AF sensors (typically about 60) were available. While 60 sensors sounds like many to choose from, it doesn’t have the resolution to do any meaningful pattern recognition by methods of machine learning (ML) or artificial intelligence (AI) despite the claims made by some manufacturers. The camera then had to rely on the RGB metering sensor (which has about 1K pixels) for subject recognition. Subject recognition by RGB (color) obviously does not work for subjects that are small and have similar color as the background. In my Canon and Nikon BIF guides I explain this issue in detail.

 

The phase-detect AF sensor of 1DX MK III (left) compare to the old 1DX MK II (right). The 1DX MK II used the old line sensor configuration providing only 61 sensors while the new sensor in the 1DX Mark III uses ~6um square pixels providing 199 sensors. The Nikon D5 and D6 sensors are similar to the 1DX MK III but with fewer sensors. Source: Canon

 

 

The Nikon Multi-CAM 20K found in the D5, D850 and D500 moved to a square pixel design for the AF sensor back in 2016, multiplying the number of sensors to more than 150. This gave a distinct and measurable advantage to Nikon cameras when it came to tracking BIF against varied backgrounds as I explained in my Nikon D5 review. It provided the best overall AF and tracking experience when shooting BIF and caused many respected bird photographers to move to a Nikon setup.  The Canon EOS-1D X Mark III followed this trend increasing the number of focus points to 199, I yet have to try the 1DX III in the field to evaluate how it stacks up against the D5. In the past Canon’s tracking algo’s were notoriously unstable and inconsistent for BIF photography. Throughout a decade of shooting with the 1D series camera I developed a series of techniques and tricks (please see my Canon BIF guide) which were required to tame the camera’s erratic behavior when it came to photographing BIF and improve the percentage of sharp frames.

Nikon’s AF system was a significant improvement for BIF, however the AF technology seems to have plateaued even in Nikon DSLR’s. From the spec’s, Nikon’s new D6 does not look like a major improvement over the D5, which is an excellent body given the fundamental DSLR focusing barriers that I just explained. Strangely the D6 went the opposite way reducing the total number of sensors to just 105 from 153 but the sensor are now larger and have less dead space between them. (shown below). The larger pixels will improve the low-light performance and the smaller gaps reduce chances of losing the subject, yet the sampling resolution is lower.

 

The AF outer coverage has hardly changed between the D5 and the D6. D6 has fewer AF points (105 vs. 153) but each sensor is now larger and some of the inner gaps have been eliminated. This is a modest evolutionary improvement. Source: Nikon

Nevertheless, it is going to be very difficult to do pattern recognition or subject detection with 100-200 pixels. The image below is a typical BIF frame, if we digitize it to just 200 pixels not much is left for the algorithm to work on. in this example, for effective subject recognition at least 1000 pixels are needed. The MILC cameras on the other hand can-in theory-have many thousands of focus points available for subject recognition and tracking.

 

Top: BIF (mallard against similarly colored BG). Bottom: Same image pixelated to simulate 1DX II (~70 AF points), 1DX III (~200 AF points) and Sony A9 (~1170 points). The duck is only discernible in the last image. It is safe to assume BIF subject recognition is simply not possible with the AF sensor alone in DSLR cameras.

 

 

The Sony A9 II has a combination of more than 1100 (693 PD + 425 CD) focus points available to it which eclipses the competition by an order of magnitude . Of course the camera’s CPU must be powerful enough to process all the data in real time and perform AF seamlessly at 20 fps, at the time of writing this article no camera other than the Sony A9 series has the kind of fire power required for real time pattern recognition and subject tracking.  Most MILC cameras are actually not there yet, not even close. I expect this to improve with time though. There will be a time when an entry level MILC camera will easily dominate today’s best pro DSLR’s

 

These fundamental differences make a mature MILC AF system inherently superior for complex and challenging subjects such as BIF against varied BG’s. For many other subjects the stat-of-the-art DSLR AF is already so good it, it cannot be improved much further. The images below are just examples to demonstrate the prowess of the A9 II’s subject tracking system.

Please note: These are straight out of the camera with no sharpening or enhancement applied. You need to click on each image to see it at full size. On a small monitor they may look compressed. 

 

Example of off-center AF tracking with Sony A9 II at close range, 1200mm at f/8. Click on the image to enlarge. the only way to get the off center subject in focus with the D5/1D X cameras is by luck.

 

Insane tracking at close range and at 1200mm with the bird partially out of the frame! click on the image to enlarge.

 

Another example of extreme tracking. Click on the image to enlarge

 

Another example. Click on the image to enlarge.  The killdeer is nowhere near the center of the frame and is zipping against a very busy and contrasty BG. Again the only way a DSLR can remain locked on the bird is such scene is pretty much by luck.

and these aren’t just a few lucky frames that any camera can get in a perfect day… they are repeatable

 

Next frame in the same sequence where the killdeer is almost lost in the BG. Click on the image to enlarge

 

Subject recognition and tracking means the camera will not be fooled even when the small subject is blocked by a large contrasty subject

 

 

tracking against a busy BG becomes an easy job for the A9 II

 

Example of tracking a black skimmer suddenly pulling up against a very distracting BG. A9 II, 1200mm. Click to enlarge

of course there are many more examples like these …

It’s not just AF:

However the differences in AF are not the only factor that make the Sony A9 II based system superior to the DSLR’s for BIF application. The other perhaps equally important advantage is the blackout-free EVF. Going back to the DSLR, when exposing a photo the mirror has to raise in order to reveal the sensor. When the mirror is up, no light gets to the viewfinder, it is totally black. It takes time for the mirror to rise, the shutter to open, and the mirror to return back to the lower position. This time referred to as black out time is about 40 msec in the pro DSLR’s and as long as 120 msec in consumer models. Canon used to publish this figure but they no longer provide it.  It is limited by the mass and the mechanics of the mirror assembly and cannot be made much faster. The vibrations caused by an under-damped mirror mechanism can have high order harmonics and cause softness in the photo even at fast shutter speeds. The main issue however, is the finder blackout when the mirror is up. Shooting at 12 fps for e.g. the total black out time during each second of shooting is simply 12 x 40 msec = 480 msec ~ 0.5 seconds! this means the finder is on average blacked out half of the time.

In practice the problem is not as bad is it sounds because the ability of the human brain to “delete” the dark frames and stitch the bright frames together. The constant interruptions appear as a continuous smooth animation to our eye thanks to our brains, just like watching a flip book animation. However the dark gaps start to become limiting when the photographer is hand holding a larger super-telephoto lens and trying to pan with erratic  motion at a high magnification. The interference from the finder blackout often results in less than perfect or off-center tracking of BIF, clipping the wings, missing sudden moves or interactions. Sometimes you can lose the bird not being able to see which way it is going.

Of course MILC cameras do not have a mirror blackout. However, if mechanical shutter is used there will be some EVF blackout when the shutter curtain is traveling in front of the sensor. It is thus necessary for the MILC to perform flawlessly with electronic shutter to eliminate the finder blackout.

Although the MILC can be made free of blackout if an electronic shutter is used, most MILC cameras on the market suffer from a finder lag phenomena. The image shown in the finder is not quite real time and has a slight lag to the scene. When you shoot a burst the camera feels like it is showing a slideshow of the captured frames in the EVF.  As you can imagine it is only frustrating to pan with a big lens with a laggy EVF that can stutter. It seriously hinders the application of  most MILC for intense action. At lower frame rates the DSLR blackout is actually much more manageable than the EVF lag because our brain cannot make up for the lag while panning.

The Sony A9 series cameras are the only cameras on the marker that can offer a totally lag and blackout-free EVF experience. You can hardly tell a photo is being recorded when looking through the EVF of an A9 II camera. This means the photographer can track erratic subjects such as BIF MUCH better and place the bird inside the frame even when shooting at 1200mm. The examples below, all taken at 1200mm and at closer range, would have been incredibly difficult to get with a DSLR due to the finder blackout.

Finally lack of any mirror shock or vibrations makes it even easier to track the BIF while shooting hand hold rock steady at 1200mm.

 

To summarize the points above we can see the Sony A9 II system advantage boils down to:

 

  • Light-weight and well balanced  600mm f/4 super-telephoto lens, native to and optimized for use with the A9 series from the get go (no funny adapter etc. needed). Tack sharp 1.4X and 2X TC’s optimized for the system
  • On-sensor PD and CD can use all the light available, AF performance largely independent of aperture between f/1.8 and f/8 (minus very dim light obviously)
  • Almost 100% AF coverage across the frame
  • high resolution AF (1100+ points) and powerful CPU allow for multi-factor real time subject tracking based on machine learning AI (pattern recognition). These algorithms are at work in both subject tracking and non-tracking AF modes.
  • black-out and lag free EVF enables seamless tracking  at close range, when using very long focal lengths. General tracking of challenging BIF becomes much easier.
  • combination of PDAF and CDAF  significantly reduces the ratio of in-focus but soft shots that plague many great frames with DSLR cameras. The focus is much more digital with the A9 II, either tack sharp or nothing in focus, most of the time.

 

Of course a new technology doesn’t always change the end result, while all of the technical advantages are there, if the pro DSLR AF is already adequate for the job the MILC will not really bring anything new to the table that wasn’t possible before. The optical viewfinder also has some inherent and fundamental advantages over the EVF which cannot be overlooked.

Whether the Sony A9 II is game changer or not really depends on the subject you intend to shoot.  If you are shooting large wildlife in African safari for example there is really nothing the A9 II does that the D5 cannot do, minus a scene that really requires 20 fps with only the Sony can do. Same if you are shooting a weekend kids’ game or a wedding.

But if you are after certain dynamic BIF images, especially those made hand held with a long super-telephoto lens the advantages and the possibilities the Sony A9 II system brings to table cannot be overlooked. It seems for this application the Sony MILC system is the ultimate solution with native the super telephoto lens and a powerful AF system to drive it.

As of now no other MILC camera on the market is even close to the Sony A9 II. Rolling shutter phenomena, finder lag, stutter and slide-show effect, slow and underwhelming AF and lack of native fast super-telephoto lenses are issues that hold back MILC compared to a pro DSLR like the D5 when it comes to BIF. I find the Nikon’s Z series cameras for example, suffer from all of these issues simultaneously making them totally useless for this kind of action photography.

The Sony A9 series is a unicorn in the field which surely deserves a special place given the many technical innovations needed in its making. It will take time for other to match this camera and also come out with a native MILC super-telephoto lenses..

 

Few examples of images taken with the Sony A9 II and 600 f/4 GM + 2X TC (1200mm) hand held. As I shoot more with this rig I hope to share more photos of different species

 

Black skimmer.  Sony A9 II, 600mm f/4 GM + TC 2X. ISO-2000 f/8 1/3200 sec hand held. Processed with C1P. It is very difficult to fit” the banking skimmer in the finder at 1200mm hand held if the viewfinder has a constant black out. Vertical frame cropped from a full-frame horizontal click on the image to enlarge

 

American Avocet. Full frame image captured with the Sony A9 II, 600mm f/4 GM + TC 2X. ISO-2000 f/8 1/3200 sec hand held. Processed with C1P.  click on the image to enlarge

 

Black-necked stilt in low light. A9 II and 600 f/4 GM + TC 2X. ISO-1250 f/8 at 1/4000 sec hand held. processed with C1P. Click on the image to enlarge

 

 

Black skimmer, Sony A9 II 600 f/4 GM + 2X TC ISO 1600 f/8 1/3200 sec hand held processed with C1P. Click on the image to enlarge