At the launch event, it was announced as such:
And the front camera has another role to play. Pixel 7 and 7 Pro’s front camera uses our advanced machine learning models for face recognition, to power Face Unlock. So you have the faster, more secure under-display fingerprint reader, and now you have the convenience of unlocking your phone with a glance.
Instead of using a dot projector, flood illuminator, and IR cameras in conjunction with the Soli radar, Google turned to a front-facing camera that supports DPAF (dual-pixel auto-focus) to presumably create a depth map to be used with the aforementioned ML models. I find it slightly weird that Google has yet to detail the technology behind this Face Unlock approach in an AI/Research blog post, like it did before.
For the most part, the experience has been fine. It’s not better than the dedicated hardware of the Pixel 4 or Face ID on the iPhone. That approach lets face recognition operate in all lighting conditions, including the absence of any. My usage of the fingerprint sensor definitely picks up at night, resulting in a two-tiered security experience throughout the day.
Of course, the more stark tiering is how face recognition is not considered a secure unlock method by Android for things related to payments, passwords, and passkeys. In that regard, today’s Face Unlock is less secure than what came before in an unfortunate and uncharacteristic step backward. Assuming Google decided to continue the Pixel 4 lineage rather than reorienting, the Pixel 5 would have probably not had a hole punch and instead opted for an iPhone-esque cutout in order to avoid a thick upper bezel. The design would have presumably shrunk by now. That’s one big shame with the Pixel restart.
Going back to reality, the Face Unlock we do have today on the Pixel 7, 7 Pro, 7a, and Fold is best summarized as a convenience. I’ve been using it a bit more with the Pixel Fold as a way to transition from an under-display to a side-mounted fingerprint sensor. Having used the former for two generations now, I’ve come to like how you’re literally interacting with the UI for unlocking. I’m amused at the hardware/software synergy.
In optimal conditions, Face Unlock on the Fold works quite fast and lets you start doing things on your phone pretty quickly, which is arguably what’s most important.
That said, the security and environmental conditions will always stop it from becoming ubiquitous. I wonder whether the primarily software-driven approach to Face Unlock is something that has a future. The roadmap behind the IR approach that Apple uses is more straightforward in that ideally all those components can be placed and work underneath the display.
Namely, for Google, will the recognition algorithms get foolproof enough that it cannot be tricked and gets elevated to a secure unlock method? Does this require a front-facing camera upgrade, more on-device processing, or another innovation entirely?
Or will the future of Face Unlock on the Pixel be a return to IR? That would be great from a security and convenience perspective but would make the current feature feel one-off.