As far as I can tell this does nothing but harm the user experience. TouchID is pretty much perfect. The physical home button is awesome and makes using the phone at lot more efficient. With TouchID I can take the phone out of my pocket and by the time I glimpse at it from any angle the phone is unlocked. How will this work when wearing glasses of any kind? Probably need to save those images just like different prints. Also probably need to look directly at the camera before anything happens.
I guess I can't blame them too much. They need to follow the newest fads. A while back that was phablets. Currently it's bezel-less designs. There is only so much you can do to a rectangular screen and they need to change something to get people to upgrade. Luckily for them people will buy whatever they put out, even if it's a downgrade in user experience.
People thought horses were perfect, in their day, and automobiles were fads.
The thing is, they weren't (and aren't), and the same is true for Touch ID. Sure, you like it, you trust it, it works well for your needs. You have trouble imagining better. No problem. You'll wait and see. Most likely, you were not enthusiastic about Touch ID when it was announced, either. The proof, as they say, was in the pudding.
I'd put Facial ID into a different class than bezel-less designs. The bezel-thing is driven by the notion that the phone doesn't have to be larger than the screen dimensions - either the screen can grow to match the dimensions of the phone, or the phone can shrink to match the dimensions of the display. Not such a big deal in the big scheme of things.
3D Facial ID, on the other hand, is driven by a variety of technologies that go beyond mere identification. The technology used to judge a face from various angles is the same needed for AR - to sense the position of the camera/display in relation to a 3D world. Facial ID is just the tip of the AR iceberg. AR is not just about game play, or superimposing animated characters on a screen (those may be judged to be somewhat faddish). It's also about measuring and interpreting a 3D world, rather than artificially flattening it to distorted 2D. It'll include using hand gestures to control the computer, rather than a physical mouse or trackpad. It includes photographing a room or object and immediately converting it to an accurate floor plan or engineering drawing. There's a lot of potential here.
The fundamentals of biometric ID are little different, whether we're talking about fingerprints, facial recognition, iris scan, or handwriting analysis. In all cases, the system is reading specific, identifiable points that have a geometric relationship to other points - it's a relatively abstract map, not a fully-representational photograph. On a fingerprint, iris, or signature, those points are measured in two dimensions. 3D facial recognition adds another dimension, making it potentially far more accurate. In all cases, the baseline scans taken during setup collect far more data points than are needed to positively identify the individual during a routine ID scan. If someone happens to be wearing glasses during baseline scan, there will still be plenty of useful data points available if the person is later scanned without glasses.
The superiority of one biometric method over another is not a matter of fundamentals, but of the quality of execution. Any given implementation of biometric scanning may suck, or it may be brilliant. If you think Apple's implementation of Touch ID is brilliant, is there a reason to believe its implementation of 3D Facial ID would necessarily suck?
