Two (2) Reasons Why the AF-Tracking of the X-T2 (and X-Pro2 Ver.2.00) is Much Better Than on the X-T1!

[Patrick FR]

Reason 1 for better AF-Tracking

“The new technology has been developed to allow increased number of AF search between the frames. There used to be only one AF search between the frames, but this is now increased up to 10. This will allow better tracking on the moving subject between the frames. […] longer live-view means that AF search can be conducted for longer period of time.”

Reason 2 for better AF-Tracking

“The AF search can now be processed in parallel. The breakdown of AF search is “Exposure”, “Read”, and then “Process”. The new algorithm allows so that each step can be conducted simultaneously. It means that while the first AF search is being “processed”, the second AF search starts to “read”, and the third AF search starts the “exposure”. By doing so, multiple AF search is now possible in the short interval.“

 

http://fujifilm-x.com/de/x-stories/evolution-of-autofocus-for-x-t2-and-x-pro2-ver200/

[johant]

It would be a shame if the X-T2 was not better. Newer generation, faster processor, lessons learned from previous models.

 

However, I am surprised to hear that Fujifilm did not use parallel processing in the X-T1. It sounds like a logical optimisation.

Edited August 4, 2016 by johant

[RadBadTad]

It would be a shame if the X-T2 was not better. Newer generation, faster processor, lessons learned from previous models.

 

However, I am surprised to hear that Fujifilm did not use parallel processing in the X-T1. It sounds like a logical optimisation.

Chances are good that the processor on the X-T1 simply couldn't handle it. 

[Pompo]

I hope shooting at high fps the blackout between frames will be minimal because on a recent youtube video I have seen ( without the battery grip) that blackout was causing a big problem in letting the photographer follow the subject and framing it.

[Aswald]

I hope shooting at high fps the blackout between frames will be minimal because on a recent youtube video I have seen ( without the battery grip) that blackout was causing a big problem in letting the photographer follow the subject and framing it.

 

It's like watching a 10 fps stop motion picture. You'll get used to it after a while. Initially, I ended up watching the motion picture and forgot about tracking the subject after a few bursts...

 

All in all, it tracks exceptionally well. Only anomaly I found which did not really exist (dslr) is the exposure difference between each shot within the burst. They may be about 1/3rd stop difference between shots. It's a little unusual when you view the contact sheets of all the shots.

Edited August 29, 2016 by Aswald

[Pompo]

I wonder why the difference in exposure ? 1/3 stop between two frames.... Does that mean one stop from an initial shot a 6 frame sequence to the 6th? That would be a bit too much!

[Aswald]

I wonder why the difference in exposure ? 1/3 stop between two frames.... Does that mean one stop from an initial shot a 6 frame sequence to the 6th? That would be a bit too much!

 

I'm not really sure either.

 

I'm thinking it could be down to high accuracy of the af point/exposure. Case in point, I use the smallest af square and as I tracked, say a face, the af square will hover over say, nose, mouth, eye brow, etc..... That could cause the shift in exposure, even though I personally think it shouldn't make so much difference.

 

The difference between each picture in the burst could be a 1/3rd stop different from the subsequent pic. It could be brighter or darker at random. In a series of 6 shots, maximum difference in exposure is 1/3rd stop. Like a random collection of -1/3, 0, +1/3.

 

I'll probably try to mount the camera on a tripod and test it on a static object. Will update results then.