So much of the discourse surrounding the latest cameras is usually about Eye AF: how fast it is, how consistent, is camera X’s better than camera Y’s. There’s a reason for that, and it’s not just because it’s the latest buzzword.
Instead, Eye AF represents a couple things. Eye AF is one of the clearest cases of value added by mirrorless cameras. As a software-dependent feature, it’s an opportunity for camera makers to distinguish their product lines without adding anything to the bill of materials. Lastly, and somewhat unfortunately for photographers, it represents the result of other major specs hitting the point of diminishing returns.
A Brief History of Eye AF
Focus tracking, and more specifically subject tracking autofocus is nothing new. DSLR cameras, with their phase-detect autofocus setups, were able to track across a couple of dozen points with decent accuracy. High-level professional bodies, like Nikon’s Dx or Canon’s 1D series cameras could offer something like 50 AF points, spread across the center 1/3 of the frame. Tracking within that zone was actually pretty competent, but as the area was small and the amount of information available to the AF system was limited, it typically wasn’t capable of things like eye or face-aware AF at the same level as mirrorless cameras.
With the advent of mirrorless cameras and on-sensor phase detect, these features got significantly better. Now there were phase-detect points across 90% or more of the frame, and they could number in the hundreds or thousands. Combined with the readout from the sensor itself, it was now feasible to track subjects with much more precision. Eye AF became more consistent, and the availability, or lack thereof, was discussed as a sticking point for buyers.
Face detect AF, and by extension Eye AF, required tight integration between the lens AF performance, image sensor readout and processing, and the software algorithms driving focus. Each of those points has represented an area of potential improvement, and with the last couple of generations of mirrorless cameras, manufacturers have made those improvements.
At least for me, the autofocus improvements available in mirrorless cameras were a major reason for my switch. Looking beyond Eye AF, moving focus on the chip meant no need to calibrate lenses, as well as no need to focus and recompose. I’d imagine that AF improvements were a significant factor for a number of other users, as AF has served as a major marketing point for most recent mirrorless releases — the latest speed oriented mirrorless cameras have made that even clearer, representing clear competition to the highest levels of DSLRs. For dedicated portrait, wedding, and event photographers, focus and recompose was often no longer necessary. Now it just became a case of selecting the correct eye and firing away.
Eye AF, and the associated improvements aren’t just of value to photographers, though. They also represent significant and meaningful savings to camera makers. Consider the manufacturing complexity involved in a DSLR’s AF: there are multiple mirrors to align, a whole separate sensor module, more complex power, and data routing, all underpinned by a need for tight tolerances between these parts. For a mirrorless camera, basically, all of that can be left off. A mirrorless camera, while it might have a slightly more expensive sensor owing to the inclusion of PDAF points, needs none of those additional parts or tolerance adjustments.
Furthermore, as a feature that is 50% or more software-based, it’s possible to improve every camera in the field with just a few lines of code, delivered via a firmware update. Unfortunately, this can cut the other way, with crippled firmware artificially reducing performance on lower-end bodies, or serving as a way of charging for feature additions after the initial sale (imagine an AF 2.0 upgrade, similar to the raw video output upgrade). Thankfully, that hasn’t yet happened. Instead, firmware updates now actually leave you with a better-performing camera months after release.
One unfortunate dimension to this conversation is that we seem to have hit a point of diminishing returns in a number of areas that previously served as banner features. Early on, there was the megapixel race, which mostly died down once cameras hit the 30 to 50-megapixel range. ISO competition has also petered out, with high-performing sensors delivering at least functional 6400 ISO. Dynamic range was a more recent one, but it seems Canon’s latest sensor tech has mostly caught up to Sony’s work, and most cameras now deliver great recovery. Frame rates have risen to the point of being fast enough, with even the resolution-oriented Z7 capable of frame rates that were the domain of sports cameras a few years ago.
One competition that just recently started is the race for a wider aperture. The new mirrorless mounts have enabled even faster lens designs, with things like f/2 zooms and absurd f/.95 primes. With the correspondingly razor-thin depth of field, these lenses demand better than ever AF performance, at least when they can be bothered to include AF in them. To make these designs usable, AF performance has to be great.
As a result, it’s clear that it’s in the manufacturer's interest to focus on autofocus. Not only does improved AF capability represent a carrot for those who haven’t yet switched, it’s also a helpful cost-savings measure in an industry that needs to pinch every penny. Photographers benefit from a better performing system, capable of delivering consistent results under a variety of conditions, and capable of being used with exciting new lens designs.
Things have come a long way from my first experience with live view AF, which was a clunky, contrast-detect experience. Fortunately, this progress doesn’t seem to be slowing down yet, as chip speeds are still improving, and even-faster sensors promise more data for those algorithms to crunch. So, while Eye AF might still be a bit of a buzzword, it represents a feature that actually delivers value to photographers, and shows that new cameras haven’t run out of development opportunities yet.
Lead image courtesy of Amanda Dalbjörn