Autofocus

ISO 250 | 1/200 sec | F2.8 | Panasonic S Pro 24-70 F2.8 @ 52mm
Photo by Jordan Drake

Key takeaways:

  • Panasonic's improvements to the Depth from Defocus system have reduced 'flutter', but it's still present
  • Face, eye, body and head detection work as advertised
  • The camera handles continuous AF well, and subject tracking is very good for a contrast-detect-based camera

System overview

Like other Lumix models, the S5 relies on Panasonic’s Depth from Defocus AF system, which first compares two out-of-focus images to drive the lens to an approximate distance, and then fine-tunes using contrast-detection. We’ve generally seen more refined AF from phase detect systems, but Panasonic continues to iterate on the technology.

One of the enhancements to DFD is the addition of head detection, adding to the existing face, eye and body detection. This addresses one of the downsides of face/eye detection: the camera losing the subject if they turn away from the camera. The S5 does a pretty good job of keeping a subject in focus, whether they're facing the camera, turned to the side or have an obstruction in the way of their face.

The S5 also has animal AF, which worked as advertised with the dogs and cats that are part of the DPReview family.

The other change to the AF system attempts to remedy probably the most frustrating aspect of DFD: fluttering (also sometimes described as wobble or hunting). When shooting stills, you'll notice it in the EVF, which can be distracting. The fluttering does not affect the very good responsiveness and accuracy of the autofocus system, however.

Performance

To see how the S5's continuous AF and subject tracking performed, we put it through our socially distanced bike test. First up is straight-on, with a single-point selected, to test the camera's ability to assess subject distance and then drive the focus element to the correct spot. We used the Panasonic S Pro 70-200 F2.8 lens at F2.8 at 183mm.

Despite its unconventional autofocus system, the S5 didn't struggle at all in this use case: it delivered a 100% hit rate, which is what we'd expect from all modern ILCs.

Now, let's take a look at the more challenging use case: an unpredictable moving subject. We used the generic tracking mode here (rather than head/face/eye detection); given the distance to the subject and the size of the tracking box, we basically selected all of Dan's upper body and the camera chose to lock onto the yellow vest. We'll use that as the reference point for focus accuracy.

The S5 turned in a good performance here, though to our eyes, two out of the 12 shots fall below our threshold for critical focus. That puts the S5 and its DFD autofocus a bit behind the best of its peers, but not by a lot.

It's worth mentioning that you can fine-tune the continuous AF system on the S5, with four presets available for stills. (There are two presets for videos in a separate menu.) AF sensitivity refers to how quickly the system reacts when the subject distance changes quickly, while AF area switching sensitivity relates to focus-point switching in the 225-pt mode. Moving subject prediction adjusts how quickly the camera reacts to subject movement.