Tracking with the A55

Here are my findings with the Sony A55V in 10 FPS, and the Sony A33 in 7 FPS mode. Note that due to the inability to repeat the test consistently, you cannot take these results as fact. The subjects were mostly waterskiers, with varying speed, and the camera operator (me) is not a experienced sports shooter. Hence, when I framed too tight, it caused the subject to be out of frame and out of focus.

A proper scientific test would use a subject with fixed speed and fixed path, e.g. a toy train on a fixed track with fixed tungsten light.

In 10 FPS mode, the camera will:

1) set the aperture as close as possible to F3.5, due to the phase-detect AF sensors having a virtual aperture of F3.5. This means a 50mm F1.4 will be set to F3.5… but a 70-400mm F4-5.6G SSM will be at 5.6 if the lens is at 400mm. No matter how bright the lighting is, it will never choose F8 if the brightest aperture of the lens is F5.6.
2) set the shutter speed to the reciprocal of the focal length (multiplied by crop factor) or faster. For 400mm it would be 1/400*1.5 = 1/600s. However since the A55V does not have 1/600s as a shutter speed, it goes to the next fastest shutter speed – 1/640s.
3) set the ISO sufficient to give exposure now that the aperture and shutter speed have been determined. In bright daylight, it would end up being ISO100.
4) if the shutter speed is too slow, speed it up to avoid overexposure (mostly on shorter focal lengths.)

This means that in bright sunlight, the Zeiss 24mm F2.0 will be shooting at 24mm F3.5 ISO100 1/500s (or however the meter reads.) Meanwhile, the Sony 70-400mm F4-5.6G SSM does 400mm F5.6 ISO100 1/640s. Of course if you point at the sun it would probably do 1/4000s!

I did not get to test either lens or combo indoors though. I imagine that using 10 FPS mode in low light without flash and a 50mm F1.4 will be bad since it will be at F3.5 all the time.

My findings with the A55V in 10 FPS mode are as follows:


It helps tracking greatly if the subject is not too big in the frame, around this size.


Don’t let the subject travel near the edges of the frame! The frame before this was in focus and the subject was in the same area.

Once the subject is out of frame, when it reenters, it may take 3-4 frames to reestablish focus. If anything, I would say it has to do with my technique (I am not a sports shooter.)

Avoid zooming in too much as there is a large risk of losing the subject totally. Better to zoom out a bit, to allow yourself room to follow the subject. This also makes for sharp, in-focus shots!


This is as tight a framing I could get without losing the subject.


This is too tight!

Does the EVF, showing the last picture shot when doing 10 FPS, cause me to lose the subject? No – I’ve been missing the subject previously whenever I zoomed in too much.

I shot a few 10 FPS sequences. I could put each picture in one of three categories – good focus, so-so focus, and out of focus.


This is good focus.


100% crop from the above.


This is so-so focus – not tack sharp, but not totally out of focus either.


100% crop from the above. Note that you could still get away with the above picture, but not the 100% crop!


This is out of focus – 100% crop.


Series 1 – waterskier shot with A55V and 70-400mm F4-5.6G SSM at 400mm.
Good focus: 19/34
So-so focus: 8/34
Out of focus: 7/34


50% crop from the A55V with 70-400mm F4-5.6G SSM at 400mm F5.6 ISO100 1/640s. From Series 1.


Likewise, from a few frames later.


Series 2 – waterskier shot with A55V and 70-400mm F4-5.6G SSM at 400mm.
Good focus: 15/30
So-so focus: 7/30
Out of focus: 8/30


Series 3 – people with A55V and Zeiss 24mm F2.0 SSM.
Good focus: 14/30
So-so focus: 6/30
Out of focus: 10/30


Series 4 – panning far away person with A55V and Zeiss 135mm F1.8.
Good focus: 5/6
So-so focus: 0/6
Out of focus: 1/6 (only the first shot – the other 5 shots look similiar)


Series 5 – panning much nearer person with A55V and Zeiss 135mm F1.8.

Once your subject is this close and near the edge of the frame, you might just lose focus. It is nearly outside the comfort zone of the A55’s AF!


Series 6 – waterskier shot with A33 and 70-400mm F4-5.6G SSM at 400mm and 7 FPS.
Good focus: 31/40
So-so focus: 0/40
Out of focus: 9/40


50% crop from DSC09505.JPG.


DSC09520.JPG.


I don’t remember how much this was cropped.

All of the shots of the waterskier were in good focus! Strangely, the out-of-focus shots were attributed to some of the boat frames.


Series 7 – people shot with A55V and Zeiss 24mm F2.0.
Good focus: 22/28
So-so focus: 3/28
Out of focus: 3/28

The gains from the translucent mirror technology are more apparent with moderate to slow subjects – I definitely would feel the performance when shooting events, gigs and weddings, for example. However I am not sure how much a sports shooter will benefit from this. While some would say the A55V equals the Canon EOS 7D in AF, and some would say it does not – I have tried the Canon EOS 7D myself and found the AF to be not that great (but this is probably due to operator error – first time I picked up a Nikon D3 with AF-S 300mm F2.8 VR I just got a bunch of back-focused faces.)

It would be best to wait for the birders and sports shooters to get the A55 to see how it performs in their trained hands.

I also shot a video using the Zeiss 24mm F2.0, tracking behind someone walking at a moderately fast pace (think street documentary) and it was able to keep focus on the subject while not having visible motion blur. However, when tracking behind someone running, the video started rolling (tilting left and right) due to running motion. The difference is like the old Handycams with SteadyShot but without 3-way shake cancelling. Unfortunately though, those videos were not in my Memory Stick…

Edited in 10:41 PM 12th September 2010 +0800 GMT::

Pros of A33/A55 over A560/A580:
– A55 has 10 FPS
– video recording has phase-detect AF (A560/A580 is manual focus video)
– no loud mirror slap sound
– no mirror shake
– the swivel LCD is arguably more practical and can be folded to protect the screen
– you can change all your settings in the EVF (ISO is missing from the A450/A500/A550; don’t know if it came back in the A560/A580)
– you can use the EVF in bright sunlight
– you can playback pictures in the EVF (useful for chimping in low light conditions and places where you don’t want the LCD to be glaring)
– you don’t get the ghosting from the side LEDs in the viewfinder that the A450/A500/A550 suffered from (I don’t know if they’ve fixed this on the A560/A580)
– A55V has built-in GPS unit
– A33/A55 has much bigger viewfinder
– A33/A55 always has phase detect AF in Live View; A560/A580 has phase detect AF in Quick AF LV, and contrast detect AF in main-sensor-based Focus Check LV (since CDAF is introduced, they cannot call it MF Check LV anymore.)
– you don’t get inconsistent WB from MF Check LV compared to Quick AF LV (my experience on the A550)

Cons of A33/A55 versus A560/A580:
– EVF exhibits RGB tearing in bright contrasty light
– weaker battery life
– 10 FPS mode shows last picture taken, not live feed
– smaller
– does not support battery grip
– supposedly less light and thus more sensor amplification (have to wait for DxO to get the official numbers)
– the shutter has to go down first and then open and then go down again and then open again (like the NEX-3 and NEX-5, or the A550 in MF Check LV)
– mirror absolutely must not be touched as there is no way to clean it
– if there is any backfocus you cannot tune it as easily as you can on a body where the AF unit is on the bottom of the camera
– in 10 FPS mode, the aperture is fixed to as close as F3.5 as possible (meaning a F1.4 lens will shoot at F3.5, and a F5.6 lens will shoot at F5.6)
– in video with AF, the aperture is fixed to as close as F3.5 as possible (meaning for videos in low light needing F1.4, you should switch to MF to allow the aperture to be at F1.4)
– EVF means you cannot manually focus the lens or check framing when the camera is off

Both the A33/A55 and A560/A580 have magnified view when focusing.

2 thoughts on “Tracking with the A55

Leave a Reply

Your email address will not be published.