I’ve toyed with the title of this article. Technically, this isn’t about eye-AF, it’s about a simplification of Fujifilm’s autofocus. The reason for this is I don’t think eye-AF is the complete solution to user problems, at least not entirely. It’s a great feature, and Sony’s facial preference technology (which gives preference to specific people) is the kind of technology that is helpful for some scenarios, but the real reason for the – popularity of eye-AF is the ability of the AF to decide the appropriate starting point for autofocus. In essence it’s the simplification of the AF system, and by simplification, I mean dumb it down to get better results without losing performance.
The current Fujifilm AF system is great but it requires constant customisation to get the best out of it, and most users don’t know how to do it, nor do they want to. You could argue that cameras are technical pieces of equipment, but in the land of iPads and Google, technical devices are becoming increasingly simple. For that reason, AF needs to be incredibly powerful and simple at the same time like ABS, traction control and similar technologies that are both powerful and simple.
How does Fujifilm solve this?
1. A simple everyday AF system.
The challenge with current zone is that it’s inconsistent. It’s hard to understand what to expect because it changes constantly. If I used zone, put the zone over a subject, I have no idea which point it will pick as a starting point. That is a problem, because there should be an option to force a starting point or at least know what it is.
For every day use, the AF system need to work similar to the current wide, where the centre point of the zone forms the initiation point for the continuous focus. This is how almost every DSLR works because it gives you control over what the camera is going to focus on. If the zone is the size of a face, you want to make sure it starts on the eye, not on the nostril (ignoring the whole eye AF concept). Whatever is under the center point, is what it should lock onto with the zone offered in different sizes depending on the level of speed required. Smaller faster, bigger not as fast but with broader coverage.
I’m not sure why Fujifilm chose not to initiate focus under the starting point, but it’s something I believe they need to change.
2. Simplified customisation for unique circumstances only.
The problem with the current customisation is the level of confusion it creates. If I’m shooting a model’s legs, which-of the modes do I select, because the examples don’t give a clear understanding of how to apply them. You can write articles, instructionals, create videos, but if you have to, you have already lost.
Firstly, there needs to be a one size fits mode that applies to 90% of scenarios. It needs to be spelt out as a one size fits all, clearly, so that users understand that it’s the recommended option to pick 90% of the time. Call it something like general.
This option should work for most circumstances so the only options you really need to cater for are the exceptions and these could be addressed as additions to #1 rather than completed separate AF systems under current AF system which is confusing. The options I would add to this are:
a) ignore obstacles (low / med / high or 1 / 2 / 3). This would be similar to Nikon’s “wait time between lock on to new item”
b) Suddenly appearing objects (on / off)
c) track object over busy background (on / off). This could also be used to give preference to objects in front.
This may seem contrary to Fujifilm’s current implementation but I feel the current implementation is confusing for many users.
3. Eye-AF / face-AF
Whether you incorporate face and eye-AF as an option d) to the above or not is a seperate issue, I’d personally prefer being able to assign it to a button, but perhaps the solution is to allow each of them to be assigned to a button depending on the user.
Whatever the option selected, the critical part is the algorithms for face and eye detect need to be improved substantially. My iPhone currently has more reliable face than my Fujifilm, so whether Fujifilm builds, acquires, or licenses the technology from a third party, it’s clear they have to get better very quickly.
As a starting point, face detect needs to be improved to a point where it can be used for high speed AF-C, the current implementation is useless for anything except s.tatic individuals and even then, hunting is fairly common. The extension to eye-AF in AFC is a logical progression that needs to follow, but the initial focus on face detect is important as a starting point.
I personally don’t believe eye AF is that critical but I suspect the reason most people feel it is important is because of the complexity of the current AF. Eye-AF takes a complex problem and makes it simple, you don’t have to move the AF point, the camera knows where to put the AF point.
4. Get rid of single point, zone and wide
With the combination of single, zone.and wide into a single setting like “all” and the ability to scroll between them, I’m not sure why you.would need them to be separated, so my perspective on this is to simply remove it.
5. An idiot proof auto mode
Yeah, I know what you’re there, why would you need an idiot proof auto mode with the above?
Because there are a lot of users buying expensive gear that doit want to change their settings so you want something that will adapt to the circumstances. My old D750 had something that switched between AF-C and AF-S, when it picked up movement and that’s what we are talking about here. An adaptive system that changes settings, reduces the number of AF points as it understands how fast the objects are moving. etc.
That’s my take. If you think I missed anything, let me know.